diff --git a/ehr_risk_kp.yaml b/ehr_risk_kp.yaml
index c379d72..ac3ed3d 100644
--- a/ehr_risk_kp.yaml
+++ b/ehr_risk_kp.yaml
@@ -813,7 +813,7 @@ components:
         parameters:
           fields: >-
             object.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -844,7 +844,7 @@ components:
         parameters:
           fields: >-
             subject.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -877,7 +877,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -908,7 +908,7 @@ components:
         parameters:
           fields: >-
             subject.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -941,7 +941,7 @@ components:
         parameters:
           fields: >-
             object.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -972,7 +972,7 @@ components:
         parameters:
           fields: >-
             subject.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1005,7 +1005,7 @@ components:
         parameters:
           fields: >-
             object.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1036,7 +1036,7 @@ components:
         parameters:
           fields: >-
             subject.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1079,7 +1079,7 @@ components:
         parameters:
           fields: >-
             object.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1110,7 +1110,7 @@ components:
         parameters:
           fields: >-
             subject.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1143,7 +1143,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1174,7 +1174,7 @@ components:
         parameters:
           fields: >-
             subject.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1207,7 +1207,7 @@ components:
         parameters:
           fields: >-
             object.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1238,7 +1238,7 @@ components:
         parameters:
           fields: >-
             subject.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1277,7 +1277,7 @@ components:
         parameters:
           fields: >-
             object.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1308,7 +1308,7 @@ components:
         parameters:
           fields: >-
             subject.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1345,7 +1345,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1376,7 +1376,7 @@ components:
         parameters:
           fields: >-
             subject.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1413,7 +1413,7 @@ components:
         parameters:
           fields: >-
             object.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1444,7 +1444,7 @@ components:
         parameters:
           fields: >-
             subject.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1477,7 +1477,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1508,7 +1508,7 @@ components:
         parameters:
           fields: >-
             subject.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1541,7 +1541,7 @@ components:
         parameters:
           fields: >-
             object.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1572,7 +1572,7 @@ components:
         parameters:
           fields: >-
             subject.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1611,7 +1611,7 @@ components:
         parameters:
           fields: >-
             object.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1642,7 +1642,7 @@ components:
         parameters:
           fields: >-
             subject.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1679,7 +1679,7 @@ components:
         parameters:
           fields: >-
             object.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1710,7 +1710,7 @@ components:
         parameters:
           fields: >-
             subject.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1743,7 +1743,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1774,7 +1774,7 @@ components:
         parameters:
           fields: >-
             subject.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1815,7 +1815,7 @@ components:
         parameters:
           fields: >-
             object.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1846,7 +1846,7 @@ components:
         parameters:
           fields: >-
             subject.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1883,7 +1883,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1914,7 +1914,7 @@ components:
         parameters:
           fields: >-
             subject.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1951,7 +1951,7 @@ components:
         parameters:
           fields: >-
             object.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -1982,7 +1982,7 @@ components:
         parameters:
           fields: >-
             subject.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2015,7 +2015,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2046,7 +2046,7 @@ components:
         parameters:
           fields: >-
             subject.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2079,7 +2079,7 @@ components:
         parameters:
           fields: >-
             object.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2110,7 +2110,7 @@ components:
         parameters:
           fields: >-
             subject.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2143,7 +2143,7 @@ components:
         parameters:
           fields: >-
             object.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2174,7 +2174,7 @@ components:
         parameters:
           fields: >-
             subject.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2211,7 +2211,7 @@ components:
         parameters:
           fields: >-
             object.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2242,7 +2242,7 @@ components:
         parameters:
           fields: >-
             subject.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2277,7 +2277,7 @@ components:
         parameters:
           fields: >-
             object.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2308,7 +2308,7 @@ components:
         parameters:
           fields: >-
             subject.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2341,7 +2341,7 @@ components:
         parameters:
           fields: >-
             object.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2372,7 +2372,7 @@ components:
         parameters:
           fields: >-
             subject.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2405,7 +2405,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2436,7 +2436,7 @@ components:
         parameters:
           fields: >-
             subject.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2469,7 +2469,7 @@ components:
         parameters:
           fields: >-
             object.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2500,7 +2500,7 @@ components:
         parameters:
           fields: >-
             subject.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2533,7 +2533,7 @@ components:
         parameters:
           fields: >-
             object.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2564,7 +2564,7 @@ components:
         parameters:
           fields: >-
             subject.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2601,7 +2601,7 @@ components:
         parameters:
           fields: >-
             object.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2632,7 +2632,7 @@ components:
         parameters:
           fields: >-
             subject.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2669,7 +2669,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2700,7 +2700,7 @@ components:
         parameters:
           fields: >-
             subject.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2737,7 +2737,7 @@ components:
         parameters:
           fields: >-
             object.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2768,7 +2768,7 @@ components:
         parameters:
           fields: >-
             subject.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2801,7 +2801,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2832,7 +2832,7 @@ components:
         parameters:
           fields: >-
             subject.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2865,7 +2865,7 @@ components:
         parameters:
           fields: >-
             object.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2896,7 +2896,7 @@ components:
         parameters:
           fields: >-
             subject.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2929,7 +2929,7 @@ components:
         parameters:
           fields: >-
             object.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2960,7 +2960,7 @@ components:
         parameters:
           fields: >-
             subject.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -2995,7 +2995,7 @@ components:
         parameters:
           fields: >-
             object.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3026,7 +3026,7 @@ components:
         parameters:
           fields: >-
             subject.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3059,7 +3059,7 @@ components:
         parameters:
           fields: >-
             object.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3090,7 +3090,7 @@ components:
         parameters:
           fields: >-
             subject.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3125,7 +3125,7 @@ components:
         parameters:
           fields: >-
             object.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3156,7 +3156,7 @@ components:
         parameters:
           fields: >-
             subject.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3189,7 +3189,7 @@ components:
         parameters:
           fields: >-
             object.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3220,7 +3220,7 @@ components:
         parameters:
           fields: >-
             subject.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3253,7 +3253,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3284,7 +3284,7 @@ components:
         parameters:
           fields: >-
             subject.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3317,7 +3317,7 @@ components:
         parameters:
           fields: >-
             object.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3348,7 +3348,7 @@ components:
         parameters:
           fields: >-
             subject.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3381,7 +3381,7 @@ components:
         parameters:
           fields: >-
             object.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3412,7 +3412,7 @@ components:
         parameters:
           fields: >-
             subject.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3445,7 +3445,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3476,7 +3476,7 @@ components:
         parameters:
           fields: >-
             subject.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3511,7 +3511,7 @@ components:
         parameters:
           fields: >-
             object.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3542,7 +3542,7 @@ components:
         parameters:
           fields: >-
             subject.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3579,7 +3579,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3610,7 +3610,7 @@ components:
         parameters:
           fields: >-
             subject.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3647,7 +3647,7 @@ components:
         parameters:
           fields: >-
             object.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3678,7 +3678,7 @@ components:
         parameters:
           fields: >-
             subject.CHEBI,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3711,7 +3711,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3742,7 +3742,7 @@ components:
         parameters:
           fields: >-
             subject.CHEBI,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3775,7 +3775,7 @@ components:
         parameters:
           fields: >-
             object.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3806,7 +3806,7 @@ components:
         parameters:
           fields: >-
             subject.CHEBI,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3839,7 +3839,7 @@ components:
         parameters:
           fields: >-
             object.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3870,7 +3870,7 @@ components:
         parameters:
           fields: >-
             subject.UNII,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3903,7 +3903,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3934,7 +3934,7 @@ components:
         parameters:
           fields: >-
             subject.UNII,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3967,7 +3967,7 @@ components:
         parameters:
           fields: >-
             object.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -3998,7 +3998,7 @@ components:
         parameters:
           fields: >-
             subject.UNII,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4031,7 +4031,7 @@ components:
         parameters:
           fields: >-
             object.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4062,7 +4062,7 @@ components:
         parameters:
           fields: >-
             subject.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4097,7 +4097,7 @@ components:
         parameters:
           fields: >-
             object.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4128,7 +4128,7 @@ components:
         parameters:
           fields: >-
             subject.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4161,7 +4161,7 @@ components:
         parameters:
           fields: >-
             object.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4192,7 +4192,7 @@ components:
         parameters:
           fields: >-
             subject.CHEBI,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4225,7 +4225,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4256,7 +4256,7 @@ components:
         parameters:
           fields: >-
             subject.CHEBI,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4289,7 +4289,7 @@ components:
         parameters:
           fields: >-
             object.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4320,7 +4320,7 @@ components:
         parameters:
           fields: >-
             subject.CHEBI,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4353,7 +4353,7 @@ components:
         parameters:
           fields: >-
             object.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4384,7 +4384,7 @@ components:
         parameters:
           fields: >-
             subject.UNII,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4417,7 +4417,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4448,7 +4448,7 @@ components:
         parameters:
           fields: >-
             subject.UNII,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4483,7 +4483,7 @@ components:
         parameters:
           fields: >-
             object.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4514,7 +4514,7 @@ components:
         parameters:
           fields: >-
             subject.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4551,7 +4551,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4582,7 +4582,7 @@ components:
         parameters:
           fields: >-
             subject.CHEBI,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4615,7 +4615,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4646,7 +4646,7 @@ components:
         parameters:
           fields: >-
             subject.UNII,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4681,7 +4681,7 @@ components:
         parameters:
           fields: >-
             object.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4712,7 +4712,7 @@ components:
         parameters:
           fields: >-
             subject.CHEBI,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4745,7 +4745,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4776,7 +4776,7 @@ components:
         parameters:
           fields: >-
             subject.CHEBI,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4809,7 +4809,7 @@ components:
         parameters:
           fields: >-
             object.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4840,7 +4840,7 @@ components:
         parameters:
           fields: >-
             subject.CHEBI,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4873,7 +4873,7 @@ components:
         parameters:
           fields: >-
             object.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4904,7 +4904,7 @@ components:
         parameters:
           fields: >-
             subject.UNII,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4939,7 +4939,7 @@ components:
         parameters:
           fields: >-
             object.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -4970,7 +4970,7 @@ components:
         parameters:
           fields: >-
             subject.UNII,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5003,7 +5003,7 @@ components:
         parameters:
           fields: >-
             object.MONDO,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5034,7 +5034,7 @@ components:
         parameters:
           fields: >-
             subject.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5069,7 +5069,7 @@ components:
         parameters:
           fields: >-
             object.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5100,7 +5100,7 @@ components:
         parameters:
           fields: >-
             subject.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5133,7 +5133,7 @@ components:
         parameters:
           fields: >-
             object.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5164,7 +5164,7 @@ components:
         parameters:
           fields: >-
             subject.CHEBI,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5197,7 +5197,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5228,7 +5228,7 @@ components:
         parameters:
           fields: >-
             subject.CHEBI,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5261,7 +5261,7 @@ components:
         parameters:
           fields: >-
             object.SNOMEDCT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5292,7 +5292,7 @@ components:
         parameters:
           fields: >-
             subject.CHEBI,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5325,7 +5325,7 @@ components:
         parameters:
           fields: >-
             object.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5356,7 +5356,7 @@ components:
         parameters:
           fields: >-
             subject.UNII,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5389,7 +5389,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5420,7 +5420,7 @@ components:
         parameters:
           fields: >-
             subject.UNII,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5455,7 +5455,7 @@ components:
         parameters:
           fields: >-
             object.HP,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5486,7 +5486,7 @@ components:
         parameters:
           fields: >-
             subject.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5519,7 +5519,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5550,7 +5550,7 @@ components:
         parameters:
           fields: >-
             subject.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5585,7 +5585,7 @@ components:
         parameters:
           fields: >-
             object.NCIT,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5616,7 +5616,7 @@ components:
         parameters:
           fields: >-
             subject.CHEBI,
-            association.edge_attributes,
+            association.edge_attributes,source.edge_sources,
             subject.name,object.name
           size: 1000
         outputs:
@@ -5641,58 +5641,70 @@ components:
       input_name: subject.name
       output_name: object.name
       edge-attributes: association.edge_attributes
+      trapi_sources: source.edge_sources
     object-NCIT:
       NCIT: object.NCIT
       input_name: subject.name
       output_name: object.name
       edge-attributes: association.edge_attributes
+      trapi_sources: source.edge_sources
     object-SNOMEDCT:
       SNOMEDCT: object.SNOMEDCT
       input_name: subject.name
       output_name: object.name
       edge-attributes: association.edge_attributes
+      trapi_sources: source.edge_sources
     object-HP:
       HP: object.HP
       input_name: subject.name
       output_name: object.name
       edge-attributes: association.edge_attributes
+      trapi_sources: source.edge_sources
     object-CHEBI:
       CHEBI: object.CHEBI
       input_name: subject.name
       output_name: object.name
       edge-attributes: association.edge_attributes
+      trapi_sources: source.edge_sources
     object-UNII:
       UNII: object.UNII
       input_name: subject.name
       output_name: object.name
       edge-attributes: association.edge_attributes
+      trapi_sources: source.edge_sources
     subject-MONDO:
       MONDO: subject.MONDO
       input_name: object.name
       output_name: subject.name
       edge-attributes: association.edge_attributes
+      trapi_sources: source.edge_sources
     subject-NCIT:
       NCIT: subject.NCIT
       input_name: object.name
       output_name: subject.name
       edge-attributes: association.edge_attributes
+      trapi_sources: source.edge_sources
     subject-SNOMEDCT:
       SNOMEDCT: subject.SNOMEDCT
       input_name: object.name
       output_name: subject.name
       edge-attributes: association.edge_attributes
+      trapi_sources: source.edge_sources
     subject-HP:
       HP: subject.HP
       input_name: object.name
       output_name: subject.name
       edge-attributes: association.edge_attributes
+      trapi_sources: source.edge_sources
     subject-CHEBI:
       CHEBI: subject.CHEBI
       input_name: object.name
       output_name: subject.name
       edge-attributes: association.edge_attributes
+      trapi_sources: source.edge_sources
     subject-UNII:
       UNII: subject.UNII
       input_name: object.name
       output_name: subject.name
-      edge-attributes: association.edge_attributes
\ No newline at end of file
+      edge-attributes: association.edge_attributes
+      trapi_sources: source.edge_sources
diff --git a/legacy_parsers/EHR_Risk_parser.ipynb b/legacy_parsers/EHR_Risk_parser.ipynb
new file mode 100644
index 0000000..6e1564d
--- /dev/null
+++ b/legacy_parsers/EHR_Risk_parser.ipynb
@@ -0,0 +1,12376 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "4d2996ec",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# qualifiers to include: https://github.com/biolink/biolink-model/issues/1050"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "0dc9bbef",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import pandas as pd\n",
+    "import json\n",
+    "import sys, os\n",
+    "import numpy as np"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "92478b80",
+   "metadata": {
+    "scrolled": true
+   },
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>id</th>\n",
+       "      <th>name</th>\n",
+       "      <th>category</th>\n",
+       "      <th>xref</th>\n",
+       "      <th>provided_by</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>HP:0025181</td>\n",
+       "      <td>Abdominal aseptic abscess</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>HP:0002027</td>\n",
+       "      <td>Abdominal pain</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>HP:0003115</td>\n",
+       "      <td>Abnormal EKG</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>HP:0006919</td>\n",
+       "      <td>Abnormal aggressive, impulsive or violent beha...</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>HP:0003119</td>\n",
+       "      <td>Abnormal circulating lipid concentration</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>801</th>\n",
+       "      <td>CHEBI:27300</td>\n",
+       "      <td>vitamin d</td>\n",
+       "      <td>biolink:ChemicalSubstance</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>802</th>\n",
+       "      <td>CHEBI:33234</td>\n",
+       "      <td>vitamin e</td>\n",
+       "      <td>biolink:ChemicalSubstance</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>803</th>\n",
+       "      <td>CHEBI:87732</td>\n",
+       "      <td>warfarin</td>\n",
+       "      <td>biolink:ChemicalSubstance</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>804</th>\n",
+       "      <td>CHEBI:36560</td>\n",
+       "      <td>zinc oxide</td>\n",
+       "      <td>biolink:ChemicalSubstance</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>805</th>\n",
+       "      <td>CHEBI:10125</td>\n",
+       "      <td>zolpidem</td>\n",
+       "      <td>biolink:ChemicalSubstance</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>806 rows × 5 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "              id                                               name  \\\n",
+       "0     HP:0025181                          Abdominal aseptic abscess   \n",
+       "1     HP:0002027                                     Abdominal pain   \n",
+       "2     HP:0003115                                       Abnormal EKG   \n",
+       "3     HP:0006919  Abnormal aggressive, impulsive or violent beha...   \n",
+       "4     HP:0003119           Abnormal circulating lipid concentration   \n",
+       "..           ...                                                ...   \n",
+       "801  CHEBI:27300                                          vitamin d   \n",
+       "802  CHEBI:33234                                          vitamin e   \n",
+       "803  CHEBI:87732                                           warfarin   \n",
+       "804  CHEBI:36560                                         zinc oxide   \n",
+       "805  CHEBI:10125                                           zolpidem   \n",
+       "\n",
+       "                      category  xref                     provided_by  \n",
+       "0    biolink:PhenotypicFeature   NaN  EHR Risk Provider (Multiomics)  \n",
+       "1    biolink:PhenotypicFeature   NaN  EHR Risk Provider (Multiomics)  \n",
+       "2    biolink:PhenotypicFeature   NaN  EHR Risk Provider (Multiomics)  \n",
+       "3    biolink:PhenotypicFeature   NaN  EHR Risk Provider (Multiomics)  \n",
+       "4    biolink:PhenotypicFeature   NaN  EHR Risk Provider (Multiomics)  \n",
+       "..                         ...   ...                             ...  \n",
+       "801  biolink:ChemicalSubstance   NaN  EHR Risk Provider (Multiomics)  \n",
+       "802  biolink:ChemicalSubstance   NaN  EHR Risk Provider (Multiomics)  \n",
+       "803  biolink:ChemicalSubstance   NaN  EHR Risk Provider (Multiomics)  \n",
+       "804  biolink:ChemicalSubstance   NaN  EHR Risk Provider (Multiomics)  \n",
+       "805  biolink:ChemicalSubstance   NaN  EHR Risk Provider (Multiomics)  \n",
+       "\n",
+       "[806 rows x 5 columns]"
+      ]
+     },
+     "execution_count": 3,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# nodes_filepath = os.path.join(\"../../data\", \"ehr_risk_nodes_data_2023_03_09.csv\")\n",
+    "# edges_filepath = os.path.join(\"../../data\", \"ehr_risk_edges_data_2023_03_09.csv\")\n",
+    "nodes_filepath = os.path.join(\"../../data\", \"ehr_risk_nodes_data_2022_06_01.csv\")\n",
+    "edges_filepath = os.path.join(\"../../data\", \"ehr_risk_edges_data_2022_06_01.csv\")\n",
+    "nodes_data = pd.read_csv(nodes_filepath, sep = ',')\n",
+    "edges_data = pd.read_csv(edges_filepath, sep = ',')\n",
+    "\n",
+    "nodes_data"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "0cb5c158",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>id</th>\n",
+       "      <th>name</th>\n",
+       "      <th>category</th>\n",
+       "      <th>xref</th>\n",
+       "      <th>provided_by</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>511</th>\n",
+       "      <td>MONDO:0009421</td>\n",
+       "      <td>hypogonadism, male</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                id                name         category  xref  \\\n",
+       "511  MONDO:0009421  hypogonadism, male  biolink:Disease   NaN   \n",
+       "\n",
+       "                        provided_by  \n",
+       "511  EHR Risk Provider (Multiomics)  "
+      ]
+     },
+     "execution_count": 3,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "nodes_data.loc[nodes_data['name'].isin(['hypogonadism, male'])]"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "0a92a7a5",
+   "metadata": {},
+   "source": [
+    "# There appears to be duplicate CURIEs \n",
+    "## This should be fixed at the enclave level"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "d9673bb6",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>id</th>\n",
+       "      <th>name</th>\n",
+       "      <th>category</th>\n",
+       "      <th>xref</th>\n",
+       "      <th>provided_by</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>HP:0025181</td>\n",
+       "      <td>Abdominal aseptic abscess</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>HP:0002027</td>\n",
+       "      <td>Abdominal pain</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>HP:0003115</td>\n",
+       "      <td>Abnormal EKG</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>HP:0006919</td>\n",
+       "      <td>Abnormal aggressive, impulsive or violent beha...</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>HP:0003119</td>\n",
+       "      <td>Abnormal circulating lipid concentration</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>801</th>\n",
+       "      <td>CHEBI:27300</td>\n",
+       "      <td>vitamin d</td>\n",
+       "      <td>biolink:ChemicalSubstance</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>802</th>\n",
+       "      <td>CHEBI:33234</td>\n",
+       "      <td>vitamin e</td>\n",
+       "      <td>biolink:ChemicalSubstance</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>803</th>\n",
+       "      <td>CHEBI:87732</td>\n",
+       "      <td>warfarin</td>\n",
+       "      <td>biolink:ChemicalSubstance</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>804</th>\n",
+       "      <td>CHEBI:36560</td>\n",
+       "      <td>zinc oxide</td>\n",
+       "      <td>biolink:ChemicalSubstance</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>805</th>\n",
+       "      <td>CHEBI:10125</td>\n",
+       "      <td>zolpidem</td>\n",
+       "      <td>biolink:ChemicalSubstance</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>804 rows × 5 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "              id                                               name  \\\n",
+       "0     HP:0025181                          Abdominal aseptic abscess   \n",
+       "1     HP:0002027                                     Abdominal pain   \n",
+       "2     HP:0003115                                       Abnormal EKG   \n",
+       "3     HP:0006919  Abnormal aggressive, impulsive or violent beha...   \n",
+       "4     HP:0003119           Abnormal circulating lipid concentration   \n",
+       "..           ...                                                ...   \n",
+       "801  CHEBI:27300                                          vitamin d   \n",
+       "802  CHEBI:33234                                          vitamin e   \n",
+       "803  CHEBI:87732                                           warfarin   \n",
+       "804  CHEBI:36560                                         zinc oxide   \n",
+       "805  CHEBI:10125                                           zolpidem   \n",
+       "\n",
+       "                      category  xref                     provided_by  \n",
+       "0    biolink:PhenotypicFeature   NaN  EHR Risk Provider (Multiomics)  \n",
+       "1    biolink:PhenotypicFeature   NaN  EHR Risk Provider (Multiomics)  \n",
+       "2    biolink:PhenotypicFeature   NaN  EHR Risk Provider (Multiomics)  \n",
+       "3    biolink:PhenotypicFeature   NaN  EHR Risk Provider (Multiomics)  \n",
+       "4    biolink:PhenotypicFeature   NaN  EHR Risk Provider (Multiomics)  \n",
+       "..                         ...   ...                             ...  \n",
+       "801  biolink:ChemicalSubstance   NaN  EHR Risk Provider (Multiomics)  \n",
+       "802  biolink:ChemicalSubstance   NaN  EHR Risk Provider (Multiomics)  \n",
+       "803  biolink:ChemicalSubstance   NaN  EHR Risk Provider (Multiomics)  \n",
+       "804  biolink:ChemicalSubstance   NaN  EHR Risk Provider (Multiomics)  \n",
+       "805  biolink:ChemicalSubstance   NaN  EHR Risk Provider (Multiomics)  \n",
+       "\n",
+       "[804 rows x 5 columns]"
+      ]
+     },
+     "execution_count": 5,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# there appears to be multiple CURIEs given for a node of the same name. Get just 1 CURIE for each name\n",
+    "nodes_data = nodes_data.drop_duplicates(subset='id', keep=\"first\")\n",
+    "nodes_data"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "id": "3eec3454",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>id</th>\n",
+       "      <th>name</th>\n",
+       "      <th>category</th>\n",
+       "      <th>xref</th>\n",
+       "      <th>provided_by</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>HP:0025181</td>\n",
+       "      <td>Abdominal aseptic abscess</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>HP:0002027</td>\n",
+       "      <td>Abdominal pain</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>HP:0003115</td>\n",
+       "      <td>Abnormal EKG</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>HP:0006919</td>\n",
+       "      <td>Abnormal aggressive, impulsive or violent beha...</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>HP:0003119</td>\n",
+       "      <td>Abnormal circulating lipid concentration</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>801</th>\n",
+       "      <td>CHEBI:27300</td>\n",
+       "      <td>vitamin d</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>802</th>\n",
+       "      <td>CHEBI:33234</td>\n",
+       "      <td>vitamin e</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>803</th>\n",
+       "      <td>CHEBI:87732</td>\n",
+       "      <td>warfarin</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>804</th>\n",
+       "      <td>CHEBI:36560</td>\n",
+       "      <td>zinc oxide</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>805</th>\n",
+       "      <td>CHEBI:10125</td>\n",
+       "      <td>zolpidem</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>804 rows × 5 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "              id                                               name  \\\n",
+       "0     HP:0025181                          Abdominal aseptic abscess   \n",
+       "1     HP:0002027                                     Abdominal pain   \n",
+       "2     HP:0003115                                       Abnormal EKG   \n",
+       "3     HP:0006919  Abnormal aggressive, impulsive or violent beha...   \n",
+       "4     HP:0003119           Abnormal circulating lipid concentration   \n",
+       "..           ...                                                ...   \n",
+       "801  CHEBI:27300                                          vitamin d   \n",
+       "802  CHEBI:33234                                          vitamin e   \n",
+       "803  CHEBI:87732                                           warfarin   \n",
+       "804  CHEBI:36560                                         zinc oxide   \n",
+       "805  CHEBI:10125                                           zolpidem   \n",
+       "\n",
+       "                      category  xref                     provided_by  \n",
+       "0    biolink:PhenotypicFeature   NaN  EHR Risk Provider (Multiomics)  \n",
+       "1    biolink:PhenotypicFeature   NaN  EHR Risk Provider (Multiomics)  \n",
+       "2    biolink:PhenotypicFeature   NaN  EHR Risk Provider (Multiomics)  \n",
+       "3    biolink:PhenotypicFeature   NaN  EHR Risk Provider (Multiomics)  \n",
+       "4    biolink:PhenotypicFeature   NaN  EHR Risk Provider (Multiomics)  \n",
+       "..                         ...   ...                             ...  \n",
+       "801     biolink:ChemicalEntity   NaN  EHR Risk Provider (Multiomics)  \n",
+       "802     biolink:ChemicalEntity   NaN  EHR Risk Provider (Multiomics)  \n",
+       "803     biolink:ChemicalEntity   NaN  EHR Risk Provider (Multiomics)  \n",
+       "804     biolink:ChemicalEntity   NaN  EHR Risk Provider (Multiomics)  \n",
+       "805     biolink:ChemicalEntity   NaN  EHR Risk Provider (Multiomics)  \n",
+       "\n",
+       "[804 rows x 5 columns]"
+      ]
+     },
+     "execution_count": 6,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# biolink:ChemicalSubstance has been depecrated. Use biolink:ChemicalEntity instead (change all values ChemicalSubstance--->ChemialEntity)\n",
+    "nodes_data[\"category\"].mask(nodes_data[\"category\"] == \"biolink:ChemicalSubstance\", \"biolink:ChemicalEntity\" , inplace=True )\n",
+    "nodes_data\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "id": "af3fe28d",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "804"
+      ]
+     },
+     "execution_count": 7,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# Check if the xref column is all NaN or empty\n",
+    "# get count of null values \n",
+    "nodes_data['xref'].isna().sum()\n",
+    "\n",
+    "# since xref is empty, we won't use it"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "id": "d3bdd0e5",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>subject</th>\n",
+       "      <th>predicate</th>\n",
+       "      <th>object</th>\n",
+       "      <th>relation</th>\n",
+       "      <th>provided_by</th>\n",
+       "      <th>provided_date</th>\n",
+       "      <th>category</th>\n",
+       "      <th>classifier</th>\n",
+       "      <th>auc_roc</th>\n",
+       "      <th>p_value</th>\n",
+       "      <th>feature_importance</th>\n",
+       "      <th>feature_coefficient</th>\n",
+       "      <th>log_positive_patient_count</th>\n",
+       "      <th>log_negative_patient_count</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>8.796399</td>\n",
+       "      <td>5</td>\n",
+       "      <td>7</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>MONDO:0010643</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.999872</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>8.585212</td>\n",
+       "      <td>5</td>\n",
+       "      <td>7</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>UNII:25ADE2236L</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.936767</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>4.558177</td>\n",
+       "      <td>5</td>\n",
+       "      <td>7</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>UNII:K16AIQ8CTM</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.998563</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>4.357522</td>\n",
+       "      <td>5</td>\n",
+       "      <td>7</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>MONDO:0007972</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.093959</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>3.926064</td>\n",
+       "      <td>5</td>\n",
+       "      <td>7</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237099</th>\n",
+       "      <td>MONDO:0013600</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.245984</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.790418</td>\n",
+       "      <td>5</td>\n",
+       "      <td>7</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237100</th>\n",
+       "      <td>MONDO:0016264</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.984249</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.796085</td>\n",
+       "      <td>5</td>\n",
+       "      <td>7</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237101</th>\n",
+       "      <td>MONDO:0004565</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.820564</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.803973</td>\n",
+       "      <td>5</td>\n",
+       "      <td>7</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237102</th>\n",
+       "      <td>CHEBI:28864</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.472603</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.822575</td>\n",
+       "      <td>5</td>\n",
+       "      <td>7</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237103</th>\n",
+       "      <td>HP:0011947</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.000207</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.825731</td>\n",
+       "      <td>5</td>\n",
+       "      <td>7</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>237104 rows × 14 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                subject                                        predicate  \\\n",
+       "0            HP:0008629  biolink:associated_with_increased_likelihood_of   \n",
+       "1         MONDO:0010643  biolink:associated_with_increased_likelihood_of   \n",
+       "2       UNII:25ADE2236L  biolink:associated_with_increased_likelihood_of   \n",
+       "3       UNII:K16AIQ8CTM  biolink:associated_with_increased_likelihood_of   \n",
+       "4         MONDO:0007972  biolink:associated_with_increased_likelihood_of   \n",
+       "...                 ...                                              ...   \n",
+       "237099    MONDO:0013600  biolink:associated_with_decreased_likelihood_of   \n",
+       "237100    MONDO:0016264  biolink:associated_with_decreased_likelihood_of   \n",
+       "237101    MONDO:0004565  biolink:associated_with_decreased_likelihood_of   \n",
+       "237102      CHEBI:28864  biolink:associated_with_decreased_likelihood_of   \n",
+       "237103       HP:0011947  biolink:associated_with_decreased_likelihood_of   \n",
+       "\n",
+       "            object    relation                     provided_by provided_date  \\\n",
+       "0       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "1       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "2       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "3       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "4       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "...            ...         ...                             ...           ...   \n",
+       "237099  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237100  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237101  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237102  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237103  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "\n",
+       "                   category           classifier   auc_roc   p_value  \\\n",
+       "0       biolink:Association  Logistic Regression  0.840132  0.000000   \n",
+       "1       biolink:Association  Logistic Regression  0.840132  0.999872   \n",
+       "2       biolink:Association  Logistic Regression  0.840132  0.936767   \n",
+       "3       biolink:Association  Logistic Regression  0.840132  0.998563   \n",
+       "4       biolink:Association  Logistic Regression  0.840132  0.093959   \n",
+       "...                     ...                  ...       ...       ...   \n",
+       "237099  biolink:Association  Logistic Regression  0.956860  0.245984   \n",
+       "237100  biolink:Association  Logistic Regression  0.956860  0.984249   \n",
+       "237101  biolink:Association  Logistic Regression  0.956860  0.820564   \n",
+       "237102  biolink:Association  Logistic Regression  0.956860  0.472603   \n",
+       "237103  biolink:Association  Logistic Regression  0.956860  0.000207   \n",
+       "\n",
+       "        feature_importance  feature_coefficient  log_positive_patient_count  \\\n",
+       "0                      NaN             8.796399                           5   \n",
+       "1                      NaN             8.585212                           5   \n",
+       "2                      NaN             4.558177                           5   \n",
+       "3                      NaN             4.357522                           5   \n",
+       "4                      NaN             3.926064                           5   \n",
+       "...                    ...                  ...                         ...   \n",
+       "237099                 NaN            -0.790418                           5   \n",
+       "237100                 NaN            -0.796085                           5   \n",
+       "237101                 NaN            -0.803973                           5   \n",
+       "237102                 NaN            -0.822575                           5   \n",
+       "237103                 NaN            -0.825731                           5   \n",
+       "\n",
+       "        log_negative_patient_count  \n",
+       "0                                7  \n",
+       "1                                7  \n",
+       "2                                7  \n",
+       "3                                7  \n",
+       "4                                7  \n",
+       "...                            ...  \n",
+       "237099                           7  \n",
+       "237100                           7  \n",
+       "237101                           7  \n",
+       "237102                           7  \n",
+       "237103                           7  \n",
+       "\n",
+       "[237104 rows x 14 columns]"
+      ]
+     },
+     "execution_count": 8,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "edges_data"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "id": "29d182f7",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "array(['biolink:associated_with_increased_likelihood_of',\n",
+       "       'biolink:associated_with_decreased_likelihood_of'], dtype=object)"
+      ]
+     },
+     "execution_count": 9,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# sanity check on predicates\n",
+    "edges_data['predicate'].unique()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "id": "2f5092be",
+   "metadata": {
+    "scrolled": true
+   },
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>subject</th>\n",
+       "      <th>predicate</th>\n",
+       "      <th>object</th>\n",
+       "      <th>relation</th>\n",
+       "      <th>provided_by</th>\n",
+       "      <th>provided_date</th>\n",
+       "      <th>category</th>\n",
+       "      <th>classifier</th>\n",
+       "      <th>auc_roc</th>\n",
+       "      <th>p_value</th>\n",
+       "      <th>feature_importance</th>\n",
+       "      <th>feature_coefficient</th>\n",
+       "      <th>num_patients_with_condition</th>\n",
+       "      <th>num_patients_without_condition</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>8.796399</td>\n",
+       "      <td>99669</td>\n",
+       "      <td>9999902</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>MONDO:0010643</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.999872</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>8.585212</td>\n",
+       "      <td>99938</td>\n",
+       "      <td>10000835</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>UNII:25ADE2236L</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.936767</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>4.558177</td>\n",
+       "      <td>99770</td>\n",
+       "      <td>10000939</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>UNII:K16AIQ8CTM</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.998563</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>4.357522</td>\n",
+       "      <td>99910</td>\n",
+       "      <td>9998659</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>MONDO:0007972</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.093959</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>3.926064</td>\n",
+       "      <td>100242</td>\n",
+       "      <td>9998750</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237099</th>\n",
+       "      <td>MONDO:0013600</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.245984</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.790418</td>\n",
+       "      <td>99922</td>\n",
+       "      <td>9995233</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237100</th>\n",
+       "      <td>MONDO:0016264</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.984249</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.796085</td>\n",
+       "      <td>99822</td>\n",
+       "      <td>10002245</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237101</th>\n",
+       "      <td>MONDO:0004565</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.820564</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.803973</td>\n",
+       "      <td>99937</td>\n",
+       "      <td>9999068</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237102</th>\n",
+       "      <td>CHEBI:28864</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.472603</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.822575</td>\n",
+       "      <td>99588</td>\n",
+       "      <td>10000517</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237103</th>\n",
+       "      <td>HP:0011947</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.000207</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.825731</td>\n",
+       "      <td>100073</td>\n",
+       "      <td>9995547</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>237104 rows × 14 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                subject                                        predicate  \\\n",
+       "0            HP:0008629  biolink:associated_with_increased_likelihood_of   \n",
+       "1         MONDO:0010643  biolink:associated_with_increased_likelihood_of   \n",
+       "2       UNII:25ADE2236L  biolink:associated_with_increased_likelihood_of   \n",
+       "3       UNII:K16AIQ8CTM  biolink:associated_with_increased_likelihood_of   \n",
+       "4         MONDO:0007972  biolink:associated_with_increased_likelihood_of   \n",
+       "...                 ...                                              ...   \n",
+       "237099    MONDO:0013600  biolink:associated_with_decreased_likelihood_of   \n",
+       "237100    MONDO:0016264  biolink:associated_with_decreased_likelihood_of   \n",
+       "237101    MONDO:0004565  biolink:associated_with_decreased_likelihood_of   \n",
+       "237102      CHEBI:28864  biolink:associated_with_decreased_likelihood_of   \n",
+       "237103       HP:0011947  biolink:associated_with_decreased_likelihood_of   \n",
+       "\n",
+       "            object    relation                     provided_by provided_date  \\\n",
+       "0       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "1       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "2       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "3       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "4       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "...            ...         ...                             ...           ...   \n",
+       "237099  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237100  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237101  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237102  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237103  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "\n",
+       "                   category           classifier   auc_roc   p_value  \\\n",
+       "0       biolink:Association  Logistic Regression  0.840132  0.000000   \n",
+       "1       biolink:Association  Logistic Regression  0.840132  0.999872   \n",
+       "2       biolink:Association  Logistic Regression  0.840132  0.936767   \n",
+       "3       biolink:Association  Logistic Regression  0.840132  0.998563   \n",
+       "4       biolink:Association  Logistic Regression  0.840132  0.093959   \n",
+       "...                     ...                  ...       ...       ...   \n",
+       "237099  biolink:Association  Logistic Regression  0.956860  0.245984   \n",
+       "237100  biolink:Association  Logistic Regression  0.956860  0.984249   \n",
+       "237101  biolink:Association  Logistic Regression  0.956860  0.820564   \n",
+       "237102  biolink:Association  Logistic Regression  0.956860  0.472603   \n",
+       "237103  biolink:Association  Logistic Regression  0.956860  0.000207   \n",
+       "\n",
+       "        feature_importance  feature_coefficient  num_patients_with_condition  \\\n",
+       "0                      NaN             8.796399                        99669   \n",
+       "1                      NaN             8.585212                        99938   \n",
+       "2                      NaN             4.558177                        99770   \n",
+       "3                      NaN             4.357522                        99910   \n",
+       "4                      NaN             3.926064                       100242   \n",
+       "...                    ...                  ...                          ...   \n",
+       "237099                 NaN            -0.790418                        99922   \n",
+       "237100                 NaN            -0.796085                        99822   \n",
+       "237101                 NaN            -0.803973                        99937   \n",
+       "237102                 NaN            -0.822575                        99588   \n",
+       "237103                 NaN            -0.825731                       100073   \n",
+       "\n",
+       "        num_patients_without_condition  \n",
+       "0                              9999902  \n",
+       "1                             10000835  \n",
+       "2                             10000939  \n",
+       "3                              9998659  \n",
+       "4                              9998750  \n",
+       "...                                ...  \n",
+       "237099                         9995233  \n",
+       "237100                        10002245  \n",
+       "237101                         9999068  \n",
+       "237102                        10000517  \n",
+       "237103                         9995547  \n",
+       "\n",
+       "[237104 rows x 14 columns]"
+      ]
+     },
+     "execution_count": 10,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "edges_data[\"num_patients_with_condition\"] = 10**(edges_data['log_positive_patient_count']) # convert log pos patient count to an actual # \n",
+    "edges_data[\"num_patients_without_condition\"] = 10**(edges_data['log_negative_patient_count']) # convert log neg patient count to an actual # \n",
+    "edges_data[\"num_patients_with_condition\"] = np.random.poisson(edges_data['num_patients_with_condition'])  # add poisson noise injection\n",
+    "edges_data[\"num_patients_without_condition\"] = np.random.poisson(edges_data['num_patients_without_condition']) # add poisson noise injection\n",
+    "edges_data = edges_data.drop(['log_positive_patient_count', 'log_negative_patient_count'], axis=1)\n",
+    "edges_data"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "id": "ef4fb895",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>subject</th>\n",
+       "      <th>predicate</th>\n",
+       "      <th>object</th>\n",
+       "      <th>relation</th>\n",
+       "      <th>provided_by</th>\n",
+       "      <th>provided_date</th>\n",
+       "      <th>category</th>\n",
+       "      <th>classifier</th>\n",
+       "      <th>auc_roc</th>\n",
+       "      <th>p_value</th>\n",
+       "      <th>feature_importance</th>\n",
+       "      <th>feature_coefficient</th>\n",
+       "      <th>num_patients_with_condition</th>\n",
+       "      <th>num_patients_without_condition</th>\n",
+       "      <th>nodes_frozenset</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>8.796399</td>\n",
+       "      <td>99669</td>\n",
+       "      <td>9999902</td>\n",
+       "      <td>(HP:0000360, HP:0008629)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>MONDO:0010643</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.999872</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>8.585212</td>\n",
+       "      <td>99938</td>\n",
+       "      <td>10000835</td>\n",
+       "      <td>(HP:0000360, MONDO:0010643)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>UNII:25ADE2236L</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.936767</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>4.558177</td>\n",
+       "      <td>99770</td>\n",
+       "      <td>10000939</td>\n",
+       "      <td>(UNII:25ADE2236L, HP:0000360)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>UNII:K16AIQ8CTM</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.998563</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>4.357522</td>\n",
+       "      <td>99910</td>\n",
+       "      <td>9998659</td>\n",
+       "      <td>(HP:0000360, UNII:K16AIQ8CTM)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>MONDO:0007972</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.093959</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>3.926064</td>\n",
+       "      <td>100242</td>\n",
+       "      <td>9998750</td>\n",
+       "      <td>(MONDO:0007972, HP:0000360)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237099</th>\n",
+       "      <td>MONDO:0013600</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.245984</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.790418</td>\n",
+       "      <td>99922</td>\n",
+       "      <td>9995233</td>\n",
+       "      <td>(MONDO:0013600, HP:0033106)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237100</th>\n",
+       "      <td>MONDO:0016264</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.984249</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.796085</td>\n",
+       "      <td>99822</td>\n",
+       "      <td>10002245</td>\n",
+       "      <td>(MONDO:0016264, HP:0033106)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237101</th>\n",
+       "      <td>MONDO:0004565</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.820564</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.803973</td>\n",
+       "      <td>99937</td>\n",
+       "      <td>9999068</td>\n",
+       "      <td>(HP:0033106, MONDO:0004565)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237102</th>\n",
+       "      <td>CHEBI:28864</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.472603</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.822575</td>\n",
+       "      <td>99588</td>\n",
+       "      <td>10000517</td>\n",
+       "      <td>(CHEBI:28864, HP:0033106)</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237103</th>\n",
+       "      <td>HP:0011947</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.000207</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.825731</td>\n",
+       "      <td>100073</td>\n",
+       "      <td>9995547</td>\n",
+       "      <td>(HP:0011947, HP:0033106)</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>237104 rows × 15 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                subject                                        predicate  \\\n",
+       "0            HP:0008629  biolink:associated_with_increased_likelihood_of   \n",
+       "1         MONDO:0010643  biolink:associated_with_increased_likelihood_of   \n",
+       "2       UNII:25ADE2236L  biolink:associated_with_increased_likelihood_of   \n",
+       "3       UNII:K16AIQ8CTM  biolink:associated_with_increased_likelihood_of   \n",
+       "4         MONDO:0007972  biolink:associated_with_increased_likelihood_of   \n",
+       "...                 ...                                              ...   \n",
+       "237099    MONDO:0013600  biolink:associated_with_decreased_likelihood_of   \n",
+       "237100    MONDO:0016264  biolink:associated_with_decreased_likelihood_of   \n",
+       "237101    MONDO:0004565  biolink:associated_with_decreased_likelihood_of   \n",
+       "237102      CHEBI:28864  biolink:associated_with_decreased_likelihood_of   \n",
+       "237103       HP:0011947  biolink:associated_with_decreased_likelihood_of   \n",
+       "\n",
+       "            object    relation                     provided_by provided_date  \\\n",
+       "0       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "1       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "2       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "3       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "4       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "...            ...         ...                             ...           ...   \n",
+       "237099  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237100  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237101  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237102  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237103  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "\n",
+       "                   category           classifier   auc_roc   p_value  \\\n",
+       "0       biolink:Association  Logistic Regression  0.840132  0.000000   \n",
+       "1       biolink:Association  Logistic Regression  0.840132  0.999872   \n",
+       "2       biolink:Association  Logistic Regression  0.840132  0.936767   \n",
+       "3       biolink:Association  Logistic Regression  0.840132  0.998563   \n",
+       "4       biolink:Association  Logistic Regression  0.840132  0.093959   \n",
+       "...                     ...                  ...       ...       ...   \n",
+       "237099  biolink:Association  Logistic Regression  0.956860  0.245984   \n",
+       "237100  biolink:Association  Logistic Regression  0.956860  0.984249   \n",
+       "237101  biolink:Association  Logistic Regression  0.956860  0.820564   \n",
+       "237102  biolink:Association  Logistic Regression  0.956860  0.472603   \n",
+       "237103  biolink:Association  Logistic Regression  0.956860  0.000207   \n",
+       "\n",
+       "        feature_importance  feature_coefficient  num_patients_with_condition  \\\n",
+       "0                      NaN             8.796399                        99669   \n",
+       "1                      NaN             8.585212                        99938   \n",
+       "2                      NaN             4.558177                        99770   \n",
+       "3                      NaN             4.357522                        99910   \n",
+       "4                      NaN             3.926064                       100242   \n",
+       "...                    ...                  ...                          ...   \n",
+       "237099                 NaN            -0.790418                        99922   \n",
+       "237100                 NaN            -0.796085                        99822   \n",
+       "237101                 NaN            -0.803973                        99937   \n",
+       "237102                 NaN            -0.822575                        99588   \n",
+       "237103                 NaN            -0.825731                       100073   \n",
+       "\n",
+       "        num_patients_without_condition                nodes_frozenset  \n",
+       "0                              9999902       (HP:0000360, HP:0008629)  \n",
+       "1                             10000835    (HP:0000360, MONDO:0010643)  \n",
+       "2                             10000939  (UNII:25ADE2236L, HP:0000360)  \n",
+       "3                              9998659  (HP:0000360, UNII:K16AIQ8CTM)  \n",
+       "4                              9998750    (MONDO:0007972, HP:0000360)  \n",
+       "...                                ...                            ...  \n",
+       "237099                         9995233    (MONDO:0013600, HP:0033106)  \n",
+       "237100                        10002245    (MONDO:0016264, HP:0033106)  \n",
+       "237101                         9999068    (HP:0033106, MONDO:0004565)  \n",
+       "237102                        10000517      (CHEBI:28864, HP:0033106)  \n",
+       "237103                         9995547       (HP:0011947, HP:0033106)  \n",
+       "\n",
+       "[237104 rows x 15 columns]"
+      ]
+     },
+     "execution_count": 11,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# Convert columns A and B to frozenset\n",
+    "# df['AB'] = df[['A', 'B']].apply(frozenset, axis=1)\n",
+    "edges_data['nodes_frozenset'] = edges_data[['subject', 'object']].apply(frozenset, axis=1)\n",
+    "edges_data"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "id": "610615b0",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>subject</th>\n",
+       "      <th>predicate</th>\n",
+       "      <th>object</th>\n",
+       "      <th>relation</th>\n",
+       "      <th>provided_by</th>\n",
+       "      <th>provided_date</th>\n",
+       "      <th>category</th>\n",
+       "      <th>classifier</th>\n",
+       "      <th>auc_roc</th>\n",
+       "      <th>p_value</th>\n",
+       "      <th>feature_importance</th>\n",
+       "      <th>feature_coefficient</th>\n",
+       "      <th>num_patients_with_condition</th>\n",
+       "      <th>num_patients_without_condition</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>8.796399</td>\n",
+       "      <td>99669</td>\n",
+       "      <td>9999902</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>MONDO:0010643</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.999872</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>8.585212</td>\n",
+       "      <td>99938</td>\n",
+       "      <td>10000835</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>UNII:25ADE2236L</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.936767</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>4.558177</td>\n",
+       "      <td>99770</td>\n",
+       "      <td>10000939</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>UNII:K16AIQ8CTM</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.998563</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>4.357522</td>\n",
+       "      <td>99910</td>\n",
+       "      <td>9998659</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>MONDO:0007972</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.093959</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>3.926064</td>\n",
+       "      <td>100242</td>\n",
+       "      <td>9998750</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237099</th>\n",
+       "      <td>MONDO:0013600</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.245984</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.790418</td>\n",
+       "      <td>99922</td>\n",
+       "      <td>9995233</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237100</th>\n",
+       "      <td>MONDO:0016264</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.984249</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.796085</td>\n",
+       "      <td>99822</td>\n",
+       "      <td>10002245</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237101</th>\n",
+       "      <td>MONDO:0004565</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.820564</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.803973</td>\n",
+       "      <td>99937</td>\n",
+       "      <td>9999068</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237102</th>\n",
+       "      <td>CHEBI:28864</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.472603</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.822575</td>\n",
+       "      <td>99588</td>\n",
+       "      <td>10000517</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237103</th>\n",
+       "      <td>HP:0011947</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033106</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.956860</td>\n",
+       "      <td>0.000207</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.825731</td>\n",
+       "      <td>100073</td>\n",
+       "      <td>9995547</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>237104 rows × 14 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                subject                                        predicate  \\\n",
+       "0            HP:0008629  biolink:associated_with_increased_likelihood_of   \n",
+       "1         MONDO:0010643  biolink:associated_with_increased_likelihood_of   \n",
+       "2       UNII:25ADE2236L  biolink:associated_with_increased_likelihood_of   \n",
+       "3       UNII:K16AIQ8CTM  biolink:associated_with_increased_likelihood_of   \n",
+       "4         MONDO:0007972  biolink:associated_with_increased_likelihood_of   \n",
+       "...                 ...                                              ...   \n",
+       "237099    MONDO:0013600  biolink:associated_with_decreased_likelihood_of   \n",
+       "237100    MONDO:0016264  biolink:associated_with_decreased_likelihood_of   \n",
+       "237101    MONDO:0004565  biolink:associated_with_decreased_likelihood_of   \n",
+       "237102      CHEBI:28864  biolink:associated_with_decreased_likelihood_of   \n",
+       "237103       HP:0011947  biolink:associated_with_decreased_likelihood_of   \n",
+       "\n",
+       "            object    relation                     provided_by provided_date  \\\n",
+       "0       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "1       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "2       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "3       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "4       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "...            ...         ...                             ...           ...   \n",
+       "237099  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237100  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237101  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237102  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237103  HP:0033106  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "\n",
+       "                   category           classifier   auc_roc   p_value  \\\n",
+       "0       biolink:Association  Logistic Regression  0.840132  0.000000   \n",
+       "1       biolink:Association  Logistic Regression  0.840132  0.999872   \n",
+       "2       biolink:Association  Logistic Regression  0.840132  0.936767   \n",
+       "3       biolink:Association  Logistic Regression  0.840132  0.998563   \n",
+       "4       biolink:Association  Logistic Regression  0.840132  0.093959   \n",
+       "...                     ...                  ...       ...       ...   \n",
+       "237099  biolink:Association  Logistic Regression  0.956860  0.245984   \n",
+       "237100  biolink:Association  Logistic Regression  0.956860  0.984249   \n",
+       "237101  biolink:Association  Logistic Regression  0.956860  0.820564   \n",
+       "237102  biolink:Association  Logistic Regression  0.956860  0.472603   \n",
+       "237103  biolink:Association  Logistic Regression  0.956860  0.000207   \n",
+       "\n",
+       "        feature_importance  feature_coefficient  num_patients_with_condition  \\\n",
+       "0                      NaN             8.796399                        99669   \n",
+       "1                      NaN             8.585212                        99938   \n",
+       "2                      NaN             4.558177                        99770   \n",
+       "3                      NaN             4.357522                        99910   \n",
+       "4                      NaN             3.926064                       100242   \n",
+       "...                    ...                  ...                          ...   \n",
+       "237099                 NaN            -0.790418                        99922   \n",
+       "237100                 NaN            -0.796085                        99822   \n",
+       "237101                 NaN            -0.803973                        99937   \n",
+       "237102                 NaN            -0.822575                        99588   \n",
+       "237103                 NaN            -0.825731                       100073   \n",
+       "\n",
+       "        num_patients_without_condition  \n",
+       "0                              9999902  \n",
+       "1                             10000835  \n",
+       "2                             10000939  \n",
+       "3                              9998659  \n",
+       "4                              9998750  \n",
+       "...                                ...  \n",
+       "237099                         9995233  \n",
+       "237100                        10002245  \n",
+       "237101                         9999068  \n",
+       "237102                        10000517  \n",
+       "237103                         9995547  \n",
+       "\n",
+       "[237104 rows x 14 columns]"
+      ]
+     },
+     "execution_count": 12,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# Find duplicate rows based on frozenset of columns A and B\n",
+    "duplicate_rows = edges_data.duplicated(['nodes_frozenset', 'auc_roc', 'p_value', 'feature_coefficient'])\n",
+    "\n",
+    "# Filter the DataFrame to remove duplicate rows\n",
+    "test = edges_data[~duplicate_rows]\n",
+    "\n",
+    "test = test.drop(columns='nodes_frozenset')\n",
+    "\n",
+    "test"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 40,
+   "id": "dc4252fb",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# # create confidence interval column by concatenating 'lower_confidence_bound'and 'upper_confidence_bound', then dropping those columns\n",
+    "# edges_data['log_odds_ratio_95_confidence_interval'] = edges_data.apply(lambda row: [row['lower_confidence_bound'], row['upper_confidence_bound']], axis=1)\n",
+    "# edges_data = edges_data.drop(['lower_confidence_bound', 'upper_confidence_bound'], axis=1)\n",
+    "# edges_data"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 13,
+   "id": "8531ffa3",
+   "metadata": {
+    "scrolled": true
+   },
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>subject</th>\n",
+       "      <th>predicate</th>\n",
+       "      <th>object</th>\n",
+       "      <th>relation</th>\n",
+       "      <th>provided_by</th>\n",
+       "      <th>provided_date</th>\n",
+       "      <th>category_x</th>\n",
+       "      <th>classifier</th>\n",
+       "      <th>auc_roc</th>\n",
+       "      <th>p_value</th>\n",
+       "      <th>feature_importance</th>\n",
+       "      <th>feature_coefficient</th>\n",
+       "      <th>num_patients_with_condition</th>\n",
+       "      <th>num_patients_without_condition</th>\n",
+       "      <th>nodes_frozenset</th>\n",
+       "      <th>id</th>\n",
+       "      <th>name</th>\n",
+       "      <th>category_y</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>8.796399</td>\n",
+       "      <td>99669</td>\n",
+       "      <td>9999902</td>\n",
+       "      <td>(HP:0000360, HP:0008629)</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0000739</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.876100</td>\n",
+       "      <td>0.985604</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.530062</td>\n",
+       "      <td>99841</td>\n",
+       "      <td>9999688</td>\n",
+       "      <td>(HP:0008629, HP:0000739)</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0000787</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.841102</td>\n",
+       "      <td>0.981654</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-1.180669</td>\n",
+       "      <td>100354</td>\n",
+       "      <td>10002196</td>\n",
+       "      <td>(HP:0008629, HP:0000787)</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0000790</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.910838</td>\n",
+       "      <td>0.960982</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-1.200019</td>\n",
+       "      <td>99835</td>\n",
+       "      <td>10001455</td>\n",
+       "      <td>(HP:0008629, HP:0000790)</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000870</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.884580</td>\n",
+       "      <td>0.988524</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>1.432611</td>\n",
+       "      <td>968</td>\n",
+       "      <td>9996545</td>\n",
+       "      <td>(HP:0008629, HP:0000870)</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237099</th>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0032312</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.921417</td>\n",
+       "      <td>0.927675</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.171607</td>\n",
+       "      <td>99721</td>\n",
+       "      <td>10001233</td>\n",
+       "      <td>(HP:0000360, HP:0032312)</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237100</th>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0032372</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.990023</td>\n",
+       "      <td>0.962012</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-2.817568</td>\n",
+       "      <td>1041</td>\n",
+       "      <td>9998402</td>\n",
+       "      <td>(HP:0000360, HP:0032372)</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237101</th>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0032473</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.963206</td>\n",
+       "      <td>0.989129</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-5.438955</td>\n",
+       "      <td>8</td>\n",
+       "      <td>10000998</td>\n",
+       "      <td>(HP:0000360, HP:0032473)</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237102</th>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033077</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.987230</td>\n",
+       "      <td>0.541701</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-1.017309</td>\n",
+       "      <td>100548</td>\n",
+       "      <td>10003263</td>\n",
+       "      <td>(HP:0000360, HP:0033077)</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237103</th>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033078</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.934115</td>\n",
+       "      <td>0.807916</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.929524</td>\n",
+       "      <td>100345</td>\n",
+       "      <td>10001251</td>\n",
+       "      <td>(HP:0000360, HP:0033078)</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>237104 rows × 18 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "           subject                                        predicate  \\\n",
+       "0       HP:0008629  biolink:associated_with_increased_likelihood_of   \n",
+       "1       HP:0008629  biolink:associated_with_decreased_likelihood_of   \n",
+       "2       HP:0008629  biolink:associated_with_decreased_likelihood_of   \n",
+       "3       HP:0008629  biolink:associated_with_decreased_likelihood_of   \n",
+       "4       HP:0008629  biolink:associated_with_increased_likelihood_of   \n",
+       "...            ...                                              ...   \n",
+       "237099  HP:0000360  biolink:associated_with_decreased_likelihood_of   \n",
+       "237100  HP:0000360  biolink:associated_with_decreased_likelihood_of   \n",
+       "237101  HP:0000360  biolink:associated_with_decreased_likelihood_of   \n",
+       "237102  HP:0000360  biolink:associated_with_decreased_likelihood_of   \n",
+       "237103  HP:0000360  biolink:associated_with_decreased_likelihood_of   \n",
+       "\n",
+       "            object    relation                     provided_by provided_date  \\\n",
+       "0       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "1       HP:0000739  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "2       HP:0000787  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "3       HP:0000790  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "4       HP:0000870  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "...            ...         ...                             ...           ...   \n",
+       "237099  HP:0032312  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237100  HP:0032372  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237101  HP:0032473  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237102  HP:0033077  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237103  HP:0033078  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "\n",
+       "                 category_x           classifier   auc_roc   p_value  \\\n",
+       "0       biolink:Association  Logistic Regression  0.840132  0.000000   \n",
+       "1       biolink:Association  Logistic Regression  0.876100  0.985604   \n",
+       "2       biolink:Association  Logistic Regression  0.841102  0.981654   \n",
+       "3       biolink:Association  Logistic Regression  0.910838  0.960982   \n",
+       "4       biolink:Association  Logistic Regression  0.884580  0.988524   \n",
+       "...                     ...                  ...       ...       ...   \n",
+       "237099  biolink:Association  Logistic Regression  0.921417  0.927675   \n",
+       "237100  biolink:Association  Logistic Regression  0.990023  0.962012   \n",
+       "237101  biolink:Association  Logistic Regression  0.963206  0.989129   \n",
+       "237102  biolink:Association  Logistic Regression  0.987230  0.541701   \n",
+       "237103  biolink:Association  Logistic Regression  0.934115  0.807916   \n",
+       "\n",
+       "        feature_importance  feature_coefficient  num_patients_with_condition  \\\n",
+       "0                      NaN             8.796399                        99669   \n",
+       "1                      NaN            -0.530062                        99841   \n",
+       "2                      NaN            -1.180669                       100354   \n",
+       "3                      NaN            -1.200019                        99835   \n",
+       "4                      NaN             1.432611                          968   \n",
+       "...                    ...                  ...                          ...   \n",
+       "237099                 NaN            -0.171607                        99721   \n",
+       "237100                 NaN            -2.817568                         1041   \n",
+       "237101                 NaN            -5.438955                            8   \n",
+       "237102                 NaN            -1.017309                       100548   \n",
+       "237103                 NaN            -0.929524                       100345   \n",
+       "\n",
+       "        num_patients_without_condition           nodes_frozenset          id  \\\n",
+       "0                              9999902  (HP:0000360, HP:0008629)  HP:0008629   \n",
+       "1                              9999688  (HP:0008629, HP:0000739)  HP:0008629   \n",
+       "2                             10002196  (HP:0008629, HP:0000787)  HP:0008629   \n",
+       "3                             10001455  (HP:0008629, HP:0000790)  HP:0008629   \n",
+       "4                              9996545  (HP:0008629, HP:0000870)  HP:0008629   \n",
+       "...                                ...                       ...         ...   \n",
+       "237099                        10001233  (HP:0000360, HP:0032312)  HP:0000360   \n",
+       "237100                         9998402  (HP:0000360, HP:0032372)  HP:0000360   \n",
+       "237101                        10000998  (HP:0000360, HP:0032473)  HP:0000360   \n",
+       "237102                        10003263  (HP:0000360, HP:0033077)  HP:0000360   \n",
+       "237103                        10001251  (HP:0000360, HP:0033078)  HP:0000360   \n",
+       "\n",
+       "                      name                 category_y  \n",
+       "0       Pulsatile tinnitus  biolink:PhenotypicFeature  \n",
+       "1       Pulsatile tinnitus  biolink:PhenotypicFeature  \n",
+       "2       Pulsatile tinnitus  biolink:PhenotypicFeature  \n",
+       "3       Pulsatile tinnitus  biolink:PhenotypicFeature  \n",
+       "4       Pulsatile tinnitus  biolink:PhenotypicFeature  \n",
+       "...                    ...                        ...  \n",
+       "237099            Tinnitus  biolink:PhenotypicFeature  \n",
+       "237100            Tinnitus  biolink:PhenotypicFeature  \n",
+       "237101            Tinnitus  biolink:PhenotypicFeature  \n",
+       "237102            Tinnitus  biolink:PhenotypicFeature  \n",
+       "237103            Tinnitus  biolink:PhenotypicFeature  \n",
+       "\n",
+       "[237104 rows x 18 columns]"
+      ]
+     },
+     "execution_count": 13,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "kg = pd.merge(edges_data, nodes_data[['id', 'name', 'category']], left_on='subject', right_on = 'id', how=\"inner\")\n",
+    "kg"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 14,
+   "id": "60ab6b78",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>subject</th>\n",
+       "      <th>predicate</th>\n",
+       "      <th>object</th>\n",
+       "      <th>relation</th>\n",
+       "      <th>provided_by</th>\n",
+       "      <th>provided_date</th>\n",
+       "      <th>predicate_category</th>\n",
+       "      <th>classifier</th>\n",
+       "      <th>auc_roc</th>\n",
+       "      <th>p_value</th>\n",
+       "      <th>feature_importance</th>\n",
+       "      <th>feature_coefficient</th>\n",
+       "      <th>num_patients_with_condition</th>\n",
+       "      <th>num_patients_without_condition</th>\n",
+       "      <th>nodes_frozenset</th>\n",
+       "      <th>subject_id</th>\n",
+       "      <th>subject_name</th>\n",
+       "      <th>subject_category</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>8.796399</td>\n",
+       "      <td>99669</td>\n",
+       "      <td>9999902</td>\n",
+       "      <td>(HP:0000360, HP:0008629)</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0000739</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.876100</td>\n",
+       "      <td>0.985604</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.530062</td>\n",
+       "      <td>99841</td>\n",
+       "      <td>9999688</td>\n",
+       "      <td>(HP:0008629, HP:0000739)</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0000787</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.841102</td>\n",
+       "      <td>0.981654</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-1.180669</td>\n",
+       "      <td>100354</td>\n",
+       "      <td>10002196</td>\n",
+       "      <td>(HP:0008629, HP:0000787)</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0000790</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.910838</td>\n",
+       "      <td>0.960982</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-1.200019</td>\n",
+       "      <td>99835</td>\n",
+       "      <td>10001455</td>\n",
+       "      <td>(HP:0008629, HP:0000790)</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000870</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.884580</td>\n",
+       "      <td>0.988524</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>1.432611</td>\n",
+       "      <td>968</td>\n",
+       "      <td>9996545</td>\n",
+       "      <td>(HP:0008629, HP:0000870)</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237099</th>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0032312</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.921417</td>\n",
+       "      <td>0.927675</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.171607</td>\n",
+       "      <td>99721</td>\n",
+       "      <td>10001233</td>\n",
+       "      <td>(HP:0000360, HP:0032312)</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237100</th>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0032372</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.990023</td>\n",
+       "      <td>0.962012</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-2.817568</td>\n",
+       "      <td>1041</td>\n",
+       "      <td>9998402</td>\n",
+       "      <td>(HP:0000360, HP:0032372)</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237101</th>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0032473</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.963206</td>\n",
+       "      <td>0.989129</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-5.438955</td>\n",
+       "      <td>8</td>\n",
+       "      <td>10000998</td>\n",
+       "      <td>(HP:0000360, HP:0032473)</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237102</th>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033077</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.987230</td>\n",
+       "      <td>0.541701</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-1.017309</td>\n",
+       "      <td>100548</td>\n",
+       "      <td>10003263</td>\n",
+       "      <td>(HP:0000360, HP:0033077)</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237103</th>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0033078</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.934115</td>\n",
+       "      <td>0.807916</td>\n",
+       "      <td>NaN</td>\n",
+       "      <td>-0.929524</td>\n",
+       "      <td>100345</td>\n",
+       "      <td>10001251</td>\n",
+       "      <td>(HP:0000360, HP:0033078)</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>237104 rows × 18 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "           subject                                        predicate  \\\n",
+       "0       HP:0008629  biolink:associated_with_increased_likelihood_of   \n",
+       "1       HP:0008629  biolink:associated_with_decreased_likelihood_of   \n",
+       "2       HP:0008629  biolink:associated_with_decreased_likelihood_of   \n",
+       "3       HP:0008629  biolink:associated_with_decreased_likelihood_of   \n",
+       "4       HP:0008629  biolink:associated_with_increased_likelihood_of   \n",
+       "...            ...                                              ...   \n",
+       "237099  HP:0000360  biolink:associated_with_decreased_likelihood_of   \n",
+       "237100  HP:0000360  biolink:associated_with_decreased_likelihood_of   \n",
+       "237101  HP:0000360  biolink:associated_with_decreased_likelihood_of   \n",
+       "237102  HP:0000360  biolink:associated_with_decreased_likelihood_of   \n",
+       "237103  HP:0000360  biolink:associated_with_decreased_likelihood_of   \n",
+       "\n",
+       "            object    relation                     provided_by provided_date  \\\n",
+       "0       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "1       HP:0000739  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "2       HP:0000787  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "3       HP:0000790  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "4       HP:0000870  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "...            ...         ...                             ...           ...   \n",
+       "237099  HP:0032312  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237100  HP:0032372  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237101  HP:0032473  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237102  HP:0033077  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237103  HP:0033078  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "\n",
+       "         predicate_category           classifier   auc_roc   p_value  \\\n",
+       "0       biolink:Association  Logistic Regression  0.840132  0.000000   \n",
+       "1       biolink:Association  Logistic Regression  0.876100  0.985604   \n",
+       "2       biolink:Association  Logistic Regression  0.841102  0.981654   \n",
+       "3       biolink:Association  Logistic Regression  0.910838  0.960982   \n",
+       "4       biolink:Association  Logistic Regression  0.884580  0.988524   \n",
+       "...                     ...                  ...       ...       ...   \n",
+       "237099  biolink:Association  Logistic Regression  0.921417  0.927675   \n",
+       "237100  biolink:Association  Logistic Regression  0.990023  0.962012   \n",
+       "237101  biolink:Association  Logistic Regression  0.963206  0.989129   \n",
+       "237102  biolink:Association  Logistic Regression  0.987230  0.541701   \n",
+       "237103  biolink:Association  Logistic Regression  0.934115  0.807916   \n",
+       "\n",
+       "        feature_importance  feature_coefficient  num_patients_with_condition  \\\n",
+       "0                      NaN             8.796399                        99669   \n",
+       "1                      NaN            -0.530062                        99841   \n",
+       "2                      NaN            -1.180669                       100354   \n",
+       "3                      NaN            -1.200019                        99835   \n",
+       "4                      NaN             1.432611                          968   \n",
+       "...                    ...                  ...                          ...   \n",
+       "237099                 NaN            -0.171607                        99721   \n",
+       "237100                 NaN            -2.817568                         1041   \n",
+       "237101                 NaN            -5.438955                            8   \n",
+       "237102                 NaN            -1.017309                       100548   \n",
+       "237103                 NaN            -0.929524                       100345   \n",
+       "\n",
+       "        num_patients_without_condition           nodes_frozenset  subject_id  \\\n",
+       "0                              9999902  (HP:0000360, HP:0008629)  HP:0008629   \n",
+       "1                              9999688  (HP:0008629, HP:0000739)  HP:0008629   \n",
+       "2                             10002196  (HP:0008629, HP:0000787)  HP:0008629   \n",
+       "3                             10001455  (HP:0008629, HP:0000790)  HP:0008629   \n",
+       "4                              9996545  (HP:0008629, HP:0000870)  HP:0008629   \n",
+       "...                                ...                       ...         ...   \n",
+       "237099                        10001233  (HP:0000360, HP:0032312)  HP:0000360   \n",
+       "237100                         9998402  (HP:0000360, HP:0032372)  HP:0000360   \n",
+       "237101                        10000998  (HP:0000360, HP:0032473)  HP:0000360   \n",
+       "237102                        10003263  (HP:0000360, HP:0033077)  HP:0000360   \n",
+       "237103                        10001251  (HP:0000360, HP:0033078)  HP:0000360   \n",
+       "\n",
+       "              subject_name           subject_category  \n",
+       "0       Pulsatile tinnitus  biolink:PhenotypicFeature  \n",
+       "1       Pulsatile tinnitus  biolink:PhenotypicFeature  \n",
+       "2       Pulsatile tinnitus  biolink:PhenotypicFeature  \n",
+       "3       Pulsatile tinnitus  biolink:PhenotypicFeature  \n",
+       "4       Pulsatile tinnitus  biolink:PhenotypicFeature  \n",
+       "...                    ...                        ...  \n",
+       "237099            Tinnitus  biolink:PhenotypicFeature  \n",
+       "237100            Tinnitus  biolink:PhenotypicFeature  \n",
+       "237101            Tinnitus  biolink:PhenotypicFeature  \n",
+       "237102            Tinnitus  biolink:PhenotypicFeature  \n",
+       "237103            Tinnitus  biolink:PhenotypicFeature  \n",
+       "\n",
+       "[237104 rows x 18 columns]"
+      ]
+     },
+     "execution_count": 14,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "kg.rename(columns = {'category_x':'predicate_category',\n",
+    "                     'category_y': 'subject_category',\n",
+    "                     'id': 'subject_id',\n",
+    "                     'name': 'subject_name'}, inplace = True)\n",
+    "kg"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 15,
+   "id": "89d414ca",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>subject</th>\n",
+       "      <th>predicate</th>\n",
+       "      <th>object</th>\n",
+       "      <th>relation</th>\n",
+       "      <th>provided_by</th>\n",
+       "      <th>provided_date</th>\n",
+       "      <th>predicate_category</th>\n",
+       "      <th>classifier</th>\n",
+       "      <th>auc_roc</th>\n",
+       "      <th>p_value</th>\n",
+       "      <th>...</th>\n",
+       "      <th>feature_coefficient</th>\n",
+       "      <th>num_patients_with_condition</th>\n",
+       "      <th>num_patients_without_condition</th>\n",
+       "      <th>nodes_frozenset</th>\n",
+       "      <th>subject_id</th>\n",
+       "      <th>subject_name</th>\n",
+       "      <th>subject_category</th>\n",
+       "      <th>id</th>\n",
+       "      <th>name</th>\n",
+       "      <th>category</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>...</td>\n",
+       "      <td>8.796399</td>\n",
+       "      <td>99669</td>\n",
+       "      <td>9999902</td>\n",
+       "      <td>(HP:0000360, HP:0008629)</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>MONDO:0010643</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.999872</td>\n",
+       "      <td>...</td>\n",
+       "      <td>8.585212</td>\n",
+       "      <td>99938</td>\n",
+       "      <td>10000835</td>\n",
+       "      <td>(HP:0000360, MONDO:0010643)</td>\n",
+       "      <td>MONDO:0010643</td>\n",
+       "      <td>acute leukemia (disease)</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>UNII:25ADE2236L</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.936767</td>\n",
+       "      <td>...</td>\n",
+       "      <td>4.558177</td>\n",
+       "      <td>99770</td>\n",
+       "      <td>10000939</td>\n",
+       "      <td>(UNII:25ADE2236L, HP:0000360)</td>\n",
+       "      <td>UNII:25ADE2236L</td>\n",
+       "      <td>thrombin</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>UNII:K16AIQ8CTM</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.998563</td>\n",
+       "      <td>...</td>\n",
+       "      <td>4.357522</td>\n",
+       "      <td>99910</td>\n",
+       "      <td>9998659</td>\n",
+       "      <td>(HP:0000360, UNII:K16AIQ8CTM)</td>\n",
+       "      <td>UNII:K16AIQ8CTM</td>\n",
+       "      <td>pertuzumab</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>MONDO:0007972</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.093959</td>\n",
+       "      <td>...</td>\n",
+       "      <td>3.926064</td>\n",
+       "      <td>100242</td>\n",
+       "      <td>9998750</td>\n",
+       "      <td>(MONDO:0007972, HP:0000360)</td>\n",
+       "      <td>MONDO:0007972</td>\n",
+       "      <td>Meniere disease</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237099</th>\n",
+       "      <td>CHEBI:114785</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.998503</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-5.087542</td>\n",
+       "      <td>1007</td>\n",
+       "      <td>10007299</td>\n",
+       "      <td>(HP:0008629, CHEBI:114785)</td>\n",
+       "      <td>CHEBI:114785</td>\n",
+       "      <td>erlotinib</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237100</th>\n",
+       "      <td>UNII:52CMI0WC3Y</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.999719</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-5.066445</td>\n",
+       "      <td>1016</td>\n",
+       "      <td>9996273</td>\n",
+       "      <td>(UNII:52CMI0WC3Y, HP:0008629)</td>\n",
+       "      <td>UNII:52CMI0WC3Y</td>\n",
+       "      <td>atezolizumab</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237101</th>\n",
+       "      <td>CHEBI:135738</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.998357</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-5.079640</td>\n",
+       "      <td>978</td>\n",
+       "      <td>9998030</td>\n",
+       "      <td>(HP:0008629, CHEBI:135738)</td>\n",
+       "      <td>CHEBI:135738</td>\n",
+       "      <td>clevidipine</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237102</th>\n",
+       "      <td>MONDO:0004967</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.997631</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-4.683547</td>\n",
+       "      <td>1006</td>\n",
+       "      <td>10001385</td>\n",
+       "      <td>(HP:0008629, MONDO:0004967)</td>\n",
+       "      <td>MONDO:0004967</td>\n",
+       "      <td>acute lymphoblastic leukemia (disease)</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237103</th>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>...</td>\n",
+       "      <td>7.553003</td>\n",
+       "      <td>997</td>\n",
+       "      <td>9997731</td>\n",
+       "      <td>(HP:0000360, HP:0008629)</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>237104 rows × 21 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                subject                                        predicate  \\\n",
+       "0            HP:0008629  biolink:associated_with_increased_likelihood_of   \n",
+       "1         MONDO:0010643  biolink:associated_with_increased_likelihood_of   \n",
+       "2       UNII:25ADE2236L  biolink:associated_with_increased_likelihood_of   \n",
+       "3       UNII:K16AIQ8CTM  biolink:associated_with_increased_likelihood_of   \n",
+       "4         MONDO:0007972  biolink:associated_with_increased_likelihood_of   \n",
+       "...                 ...                                              ...   \n",
+       "237099     CHEBI:114785  biolink:associated_with_decreased_likelihood_of   \n",
+       "237100  UNII:52CMI0WC3Y  biolink:associated_with_decreased_likelihood_of   \n",
+       "237101     CHEBI:135738  biolink:associated_with_decreased_likelihood_of   \n",
+       "237102    MONDO:0004967  biolink:associated_with_decreased_likelihood_of   \n",
+       "237103       HP:0000360  biolink:associated_with_increased_likelihood_of   \n",
+       "\n",
+       "            object    relation                     provided_by provided_date  \\\n",
+       "0       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "1       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "2       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "3       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "4       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "...            ...         ...                             ...           ...   \n",
+       "237099  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237100  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237101  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237102  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237103  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "\n",
+       "         predicate_category           classifier   auc_roc   p_value  ...  \\\n",
+       "0       biolink:Association  Logistic Regression  0.840132  0.000000  ...   \n",
+       "1       biolink:Association  Logistic Regression  0.840132  0.999872  ...   \n",
+       "2       biolink:Association  Logistic Regression  0.840132  0.936767  ...   \n",
+       "3       biolink:Association  Logistic Regression  0.840132  0.998563  ...   \n",
+       "4       biolink:Association  Logistic Regression  0.840132  0.093959  ...   \n",
+       "...                     ...                  ...       ...       ...  ...   \n",
+       "237099  biolink:Association  Logistic Regression  0.959791  0.998503  ...   \n",
+       "237100  biolink:Association  Logistic Regression  0.959791  0.999719  ...   \n",
+       "237101  biolink:Association  Logistic Regression  0.959791  0.998357  ...   \n",
+       "237102  biolink:Association  Logistic Regression  0.959791  0.997631  ...   \n",
+       "237103  biolink:Association  Logistic Regression  0.959791  0.000000  ...   \n",
+       "\n",
+       "        feature_coefficient  num_patients_with_condition  \\\n",
+       "0                  8.796399                        99669   \n",
+       "1                  8.585212                        99938   \n",
+       "2                  4.558177                        99770   \n",
+       "3                  4.357522                        99910   \n",
+       "4                  3.926064                       100242   \n",
+       "...                     ...                          ...   \n",
+       "237099            -5.087542                         1007   \n",
+       "237100            -5.066445                         1016   \n",
+       "237101            -5.079640                          978   \n",
+       "237102            -4.683547                         1006   \n",
+       "237103             7.553003                          997   \n",
+       "\n",
+       "        num_patients_without_condition                nodes_frozenset  \\\n",
+       "0                              9999902       (HP:0000360, HP:0008629)   \n",
+       "1                             10000835    (HP:0000360, MONDO:0010643)   \n",
+       "2                             10000939  (UNII:25ADE2236L, HP:0000360)   \n",
+       "3                              9998659  (HP:0000360, UNII:K16AIQ8CTM)   \n",
+       "4                              9998750    (MONDO:0007972, HP:0000360)   \n",
+       "...                                ...                            ...   \n",
+       "237099                        10007299     (HP:0008629, CHEBI:114785)   \n",
+       "237100                         9996273  (UNII:52CMI0WC3Y, HP:0008629)   \n",
+       "237101                         9998030     (HP:0008629, CHEBI:135738)   \n",
+       "237102                        10001385    (HP:0008629, MONDO:0004967)   \n",
+       "237103                         9997731       (HP:0000360, HP:0008629)   \n",
+       "\n",
+       "             subject_id                            subject_name  \\\n",
+       "0            HP:0008629                      Pulsatile tinnitus   \n",
+       "1         MONDO:0010643                acute leukemia (disease)   \n",
+       "2       UNII:25ADE2236L                                thrombin   \n",
+       "3       UNII:K16AIQ8CTM                              pertuzumab   \n",
+       "4         MONDO:0007972                         Meniere disease   \n",
+       "...                 ...                                     ...   \n",
+       "237099     CHEBI:114785                               erlotinib   \n",
+       "237100  UNII:52CMI0WC3Y                            atezolizumab   \n",
+       "237101     CHEBI:135738                             clevidipine   \n",
+       "237102    MONDO:0004967  acute lymphoblastic leukemia (disease)   \n",
+       "237103       HP:0000360                                Tinnitus   \n",
+       "\n",
+       "                 subject_category          id                name  \\\n",
+       "0       biolink:PhenotypicFeature  HP:0000360            Tinnitus   \n",
+       "1                 biolink:Disease  HP:0000360            Tinnitus   \n",
+       "2          biolink:ChemicalEntity  HP:0000360            Tinnitus   \n",
+       "3          biolink:ChemicalEntity  HP:0000360            Tinnitus   \n",
+       "4                 biolink:Disease  HP:0000360            Tinnitus   \n",
+       "...                           ...         ...                 ...   \n",
+       "237099     biolink:ChemicalEntity  HP:0008629  Pulsatile tinnitus   \n",
+       "237100     biolink:ChemicalEntity  HP:0008629  Pulsatile tinnitus   \n",
+       "237101     biolink:ChemicalEntity  HP:0008629  Pulsatile tinnitus   \n",
+       "237102            biolink:Disease  HP:0008629  Pulsatile tinnitus   \n",
+       "237103  biolink:PhenotypicFeature  HP:0008629  Pulsatile tinnitus   \n",
+       "\n",
+       "                         category  \n",
+       "0       biolink:PhenotypicFeature  \n",
+       "1       biolink:PhenotypicFeature  \n",
+       "2       biolink:PhenotypicFeature  \n",
+       "3       biolink:PhenotypicFeature  \n",
+       "4       biolink:PhenotypicFeature  \n",
+       "...                           ...  \n",
+       "237099  biolink:PhenotypicFeature  \n",
+       "237100  biolink:PhenotypicFeature  \n",
+       "237101  biolink:PhenotypicFeature  \n",
+       "237102  biolink:PhenotypicFeature  \n",
+       "237103  biolink:PhenotypicFeature  \n",
+       "\n",
+       "[237104 rows x 21 columns]"
+      ]
+     },
+     "execution_count": 15,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# merging object info from nodes df\n",
+    "kg = pd.merge(kg, nodes_data[['id', 'name', 'category']], left_on='object', right_on = 'id', how=\"inner\")\n",
+    "kg"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "id": "b81d60e9",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>subject</th>\n",
+       "      <th>predicate</th>\n",
+       "      <th>object</th>\n",
+       "      <th>relation</th>\n",
+       "      <th>provided_by</th>\n",
+       "      <th>provided_date</th>\n",
+       "      <th>predicate_category</th>\n",
+       "      <th>classifier</th>\n",
+       "      <th>auc_roc</th>\n",
+       "      <th>p_value</th>\n",
+       "      <th>...</th>\n",
+       "      <th>feature_coefficient</th>\n",
+       "      <th>num_patients_with_condition</th>\n",
+       "      <th>num_patients_without_condition</th>\n",
+       "      <th>nodes_frozenset</th>\n",
+       "      <th>subject_id</th>\n",
+       "      <th>subject_name</th>\n",
+       "      <th>subject_category</th>\n",
+       "      <th>object_id</th>\n",
+       "      <th>object_name</th>\n",
+       "      <th>object_category</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>...</td>\n",
+       "      <td>8.796399</td>\n",
+       "      <td>99669</td>\n",
+       "      <td>9999902</td>\n",
+       "      <td>(HP:0000360, HP:0008629)</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>MONDO:0010643</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.999872</td>\n",
+       "      <td>...</td>\n",
+       "      <td>8.585212</td>\n",
+       "      <td>99938</td>\n",
+       "      <td>10000835</td>\n",
+       "      <td>(HP:0000360, MONDO:0010643)</td>\n",
+       "      <td>MONDO:0010643</td>\n",
+       "      <td>acute leukemia (disease)</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>UNII:25ADE2236L</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.936767</td>\n",
+       "      <td>...</td>\n",
+       "      <td>4.558177</td>\n",
+       "      <td>99770</td>\n",
+       "      <td>10000939</td>\n",
+       "      <td>(UNII:25ADE2236L, HP:0000360)</td>\n",
+       "      <td>UNII:25ADE2236L</td>\n",
+       "      <td>thrombin</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>UNII:K16AIQ8CTM</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.998563</td>\n",
+       "      <td>...</td>\n",
+       "      <td>4.357522</td>\n",
+       "      <td>99910</td>\n",
+       "      <td>9998659</td>\n",
+       "      <td>(HP:0000360, UNII:K16AIQ8CTM)</td>\n",
+       "      <td>UNII:K16AIQ8CTM</td>\n",
+       "      <td>pertuzumab</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>MONDO:0007972</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.093959</td>\n",
+       "      <td>...</td>\n",
+       "      <td>3.926064</td>\n",
+       "      <td>100242</td>\n",
+       "      <td>9998750</td>\n",
+       "      <td>(MONDO:0007972, HP:0000360)</td>\n",
+       "      <td>MONDO:0007972</td>\n",
+       "      <td>Meniere disease</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237099</th>\n",
+       "      <td>CHEBI:114785</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.998503</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-5.087542</td>\n",
+       "      <td>1007</td>\n",
+       "      <td>10007299</td>\n",
+       "      <td>(HP:0008629, CHEBI:114785)</td>\n",
+       "      <td>CHEBI:114785</td>\n",
+       "      <td>erlotinib</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237100</th>\n",
+       "      <td>UNII:52CMI0WC3Y</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.999719</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-5.066445</td>\n",
+       "      <td>1016</td>\n",
+       "      <td>9996273</td>\n",
+       "      <td>(UNII:52CMI0WC3Y, HP:0008629)</td>\n",
+       "      <td>UNII:52CMI0WC3Y</td>\n",
+       "      <td>atezolizumab</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237101</th>\n",
+       "      <td>CHEBI:135738</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.998357</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-5.079640</td>\n",
+       "      <td>978</td>\n",
+       "      <td>9998030</td>\n",
+       "      <td>(HP:0008629, CHEBI:135738)</td>\n",
+       "      <td>CHEBI:135738</td>\n",
+       "      <td>clevidipine</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237102</th>\n",
+       "      <td>MONDO:0004967</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.997631</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-4.683547</td>\n",
+       "      <td>1006</td>\n",
+       "      <td>10001385</td>\n",
+       "      <td>(HP:0008629, MONDO:0004967)</td>\n",
+       "      <td>MONDO:0004967</td>\n",
+       "      <td>acute lymphoblastic leukemia (disease)</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237103</th>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>...</td>\n",
+       "      <td>7.553003</td>\n",
+       "      <td>997</td>\n",
+       "      <td>9997731</td>\n",
+       "      <td>(HP:0000360, HP:0008629)</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>237104 rows × 21 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                subject                                        predicate  \\\n",
+       "0            HP:0008629  biolink:associated_with_increased_likelihood_of   \n",
+       "1         MONDO:0010643  biolink:associated_with_increased_likelihood_of   \n",
+       "2       UNII:25ADE2236L  biolink:associated_with_increased_likelihood_of   \n",
+       "3       UNII:K16AIQ8CTM  biolink:associated_with_increased_likelihood_of   \n",
+       "4         MONDO:0007972  biolink:associated_with_increased_likelihood_of   \n",
+       "...                 ...                                              ...   \n",
+       "237099     CHEBI:114785  biolink:associated_with_decreased_likelihood_of   \n",
+       "237100  UNII:52CMI0WC3Y  biolink:associated_with_decreased_likelihood_of   \n",
+       "237101     CHEBI:135738  biolink:associated_with_decreased_likelihood_of   \n",
+       "237102    MONDO:0004967  biolink:associated_with_decreased_likelihood_of   \n",
+       "237103       HP:0000360  biolink:associated_with_increased_likelihood_of   \n",
+       "\n",
+       "            object    relation                     provided_by provided_date  \\\n",
+       "0       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "1       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "2       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "3       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "4       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "...            ...         ...                             ...           ...   \n",
+       "237099  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237100  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237101  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237102  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237103  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "\n",
+       "         predicate_category           classifier   auc_roc   p_value  ...  \\\n",
+       "0       biolink:Association  Logistic Regression  0.840132  0.000000  ...   \n",
+       "1       biolink:Association  Logistic Regression  0.840132  0.999872  ...   \n",
+       "2       biolink:Association  Logistic Regression  0.840132  0.936767  ...   \n",
+       "3       biolink:Association  Logistic Regression  0.840132  0.998563  ...   \n",
+       "4       biolink:Association  Logistic Regression  0.840132  0.093959  ...   \n",
+       "...                     ...                  ...       ...       ...  ...   \n",
+       "237099  biolink:Association  Logistic Regression  0.959791  0.998503  ...   \n",
+       "237100  biolink:Association  Logistic Regression  0.959791  0.999719  ...   \n",
+       "237101  biolink:Association  Logistic Regression  0.959791  0.998357  ...   \n",
+       "237102  biolink:Association  Logistic Regression  0.959791  0.997631  ...   \n",
+       "237103  biolink:Association  Logistic Regression  0.959791  0.000000  ...   \n",
+       "\n",
+       "        feature_coefficient  num_patients_with_condition  \\\n",
+       "0                  8.796399                        99669   \n",
+       "1                  8.585212                        99938   \n",
+       "2                  4.558177                        99770   \n",
+       "3                  4.357522                        99910   \n",
+       "4                  3.926064                       100242   \n",
+       "...                     ...                          ...   \n",
+       "237099            -5.087542                         1007   \n",
+       "237100            -5.066445                         1016   \n",
+       "237101            -5.079640                          978   \n",
+       "237102            -4.683547                         1006   \n",
+       "237103             7.553003                          997   \n",
+       "\n",
+       "        num_patients_without_condition                nodes_frozenset  \\\n",
+       "0                              9999902       (HP:0000360, HP:0008629)   \n",
+       "1                             10000835    (HP:0000360, MONDO:0010643)   \n",
+       "2                             10000939  (UNII:25ADE2236L, HP:0000360)   \n",
+       "3                              9998659  (HP:0000360, UNII:K16AIQ8CTM)   \n",
+       "4                              9998750    (MONDO:0007972, HP:0000360)   \n",
+       "...                                ...                            ...   \n",
+       "237099                        10007299     (HP:0008629, CHEBI:114785)   \n",
+       "237100                         9996273  (UNII:52CMI0WC3Y, HP:0008629)   \n",
+       "237101                         9998030     (HP:0008629, CHEBI:135738)   \n",
+       "237102                        10001385    (HP:0008629, MONDO:0004967)   \n",
+       "237103                         9997731       (HP:0000360, HP:0008629)   \n",
+       "\n",
+       "             subject_id                            subject_name  \\\n",
+       "0            HP:0008629                      Pulsatile tinnitus   \n",
+       "1         MONDO:0010643                acute leukemia (disease)   \n",
+       "2       UNII:25ADE2236L                                thrombin   \n",
+       "3       UNII:K16AIQ8CTM                              pertuzumab   \n",
+       "4         MONDO:0007972                         Meniere disease   \n",
+       "...                 ...                                     ...   \n",
+       "237099     CHEBI:114785                               erlotinib   \n",
+       "237100  UNII:52CMI0WC3Y                            atezolizumab   \n",
+       "237101     CHEBI:135738                             clevidipine   \n",
+       "237102    MONDO:0004967  acute lymphoblastic leukemia (disease)   \n",
+       "237103       HP:0000360                                Tinnitus   \n",
+       "\n",
+       "                 subject_category   object_id         object_name  \\\n",
+       "0       biolink:PhenotypicFeature  HP:0000360            Tinnitus   \n",
+       "1                 biolink:Disease  HP:0000360            Tinnitus   \n",
+       "2          biolink:ChemicalEntity  HP:0000360            Tinnitus   \n",
+       "3          biolink:ChemicalEntity  HP:0000360            Tinnitus   \n",
+       "4                 biolink:Disease  HP:0000360            Tinnitus   \n",
+       "...                           ...         ...                 ...   \n",
+       "237099     biolink:ChemicalEntity  HP:0008629  Pulsatile tinnitus   \n",
+       "237100     biolink:ChemicalEntity  HP:0008629  Pulsatile tinnitus   \n",
+       "237101     biolink:ChemicalEntity  HP:0008629  Pulsatile tinnitus   \n",
+       "237102            biolink:Disease  HP:0008629  Pulsatile tinnitus   \n",
+       "237103  biolink:PhenotypicFeature  HP:0008629  Pulsatile tinnitus   \n",
+       "\n",
+       "                  object_category  \n",
+       "0       biolink:PhenotypicFeature  \n",
+       "1       biolink:PhenotypicFeature  \n",
+       "2       biolink:PhenotypicFeature  \n",
+       "3       biolink:PhenotypicFeature  \n",
+       "4       biolink:PhenotypicFeature  \n",
+       "...                           ...  \n",
+       "237099  biolink:PhenotypicFeature  \n",
+       "237100  biolink:PhenotypicFeature  \n",
+       "237101  biolink:PhenotypicFeature  \n",
+       "237102  biolink:PhenotypicFeature  \n",
+       "237103  biolink:PhenotypicFeature  \n",
+       "\n",
+       "[237104 rows x 21 columns]"
+      ]
+     },
+     "execution_count": 16,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "kg.rename(columns = {'id':'object_id',\n",
+    "                     'category': 'object_category',\n",
+    "                     'name': 'object_name'}, inplace = True)\n",
+    "kg"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 17,
+   "id": "d54c7830",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>subject</th>\n",
+       "      <th>predicate</th>\n",
+       "      <th>object</th>\n",
+       "      <th>relation</th>\n",
+       "      <th>provided_by</th>\n",
+       "      <th>provided_date</th>\n",
+       "      <th>predicate_category</th>\n",
+       "      <th>classifier</th>\n",
+       "      <th>auc_roc</th>\n",
+       "      <th>p_value</th>\n",
+       "      <th>...</th>\n",
+       "      <th>feature_coefficient</th>\n",
+       "      <th>num_patients_with_condition</th>\n",
+       "      <th>num_patients_without_condition</th>\n",
+       "      <th>nodes_frozenset</th>\n",
+       "      <th>subject_id</th>\n",
+       "      <th>subject_name</th>\n",
+       "      <th>subject_category</th>\n",
+       "      <th>object_id</th>\n",
+       "      <th>object_name</th>\n",
+       "      <th>object_category</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>...</td>\n",
+       "      <td>8.796399</td>\n",
+       "      <td>99669</td>\n",
+       "      <td>9999902</td>\n",
+       "      <td>(HP:0000360, HP:0008629)</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>MONDO:0010643</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.999872</td>\n",
+       "      <td>...</td>\n",
+       "      <td>8.585212</td>\n",
+       "      <td>99938</td>\n",
+       "      <td>10000835</td>\n",
+       "      <td>(HP:0000360, MONDO:0010643)</td>\n",
+       "      <td>MONDO:0010643</td>\n",
+       "      <td>acute leukemia (disease)</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>UNII:25ADE2236L</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.936767</td>\n",
+       "      <td>...</td>\n",
+       "      <td>4.558177</td>\n",
+       "      <td>99770</td>\n",
+       "      <td>10000939</td>\n",
+       "      <td>(UNII:25ADE2236L, HP:0000360)</td>\n",
+       "      <td>UNII:25ADE2236L</td>\n",
+       "      <td>thrombin</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>UNII:K16AIQ8CTM</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.998563</td>\n",
+       "      <td>...</td>\n",
+       "      <td>4.357522</td>\n",
+       "      <td>99910</td>\n",
+       "      <td>9998659</td>\n",
+       "      <td>(HP:0000360, UNII:K16AIQ8CTM)</td>\n",
+       "      <td>UNII:K16AIQ8CTM</td>\n",
+       "      <td>pertuzumab</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>MONDO:0007972</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.093959</td>\n",
+       "      <td>...</td>\n",
+       "      <td>3.926064</td>\n",
+       "      <td>100242</td>\n",
+       "      <td>9998750</td>\n",
+       "      <td>(MONDO:0007972, HP:0000360)</td>\n",
+       "      <td>MONDO:0007972</td>\n",
+       "      <td>Meniere disease</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237099</th>\n",
+       "      <td>CHEBI:114785</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.998503</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-5.087542</td>\n",
+       "      <td>1007</td>\n",
+       "      <td>10007299</td>\n",
+       "      <td>(HP:0008629, CHEBI:114785)</td>\n",
+       "      <td>CHEBI:114785</td>\n",
+       "      <td>erlotinib</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237100</th>\n",
+       "      <td>UNII:52CMI0WC3Y</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.999719</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-5.066445</td>\n",
+       "      <td>1016</td>\n",
+       "      <td>9996273</td>\n",
+       "      <td>(UNII:52CMI0WC3Y, HP:0008629)</td>\n",
+       "      <td>UNII:52CMI0WC3Y</td>\n",
+       "      <td>atezolizumab</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237101</th>\n",
+       "      <td>CHEBI:135738</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.998357</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-5.079640</td>\n",
+       "      <td>978</td>\n",
+       "      <td>9998030</td>\n",
+       "      <td>(HP:0008629, CHEBI:135738)</td>\n",
+       "      <td>CHEBI:135738</td>\n",
+       "      <td>clevidipine</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237102</th>\n",
+       "      <td>MONDO:0004967</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.997631</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-4.683547</td>\n",
+       "      <td>1006</td>\n",
+       "      <td>10001385</td>\n",
+       "      <td>(HP:0008629, MONDO:0004967)</td>\n",
+       "      <td>MONDO:0004967</td>\n",
+       "      <td>acute lymphoblastic leukemia (disease)</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237103</th>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>...</td>\n",
+       "      <td>7.553003</td>\n",
+       "      <td>997</td>\n",
+       "      <td>9997731</td>\n",
+       "      <td>(HP:0000360, HP:0008629)</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>237104 rows × 21 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                subject                                        predicate  \\\n",
+       "0            HP:0008629  biolink:associated_with_increased_likelihood_of   \n",
+       "1         MONDO:0010643  biolink:associated_with_increased_likelihood_of   \n",
+       "2       UNII:25ADE2236L  biolink:associated_with_increased_likelihood_of   \n",
+       "3       UNII:K16AIQ8CTM  biolink:associated_with_increased_likelihood_of   \n",
+       "4         MONDO:0007972  biolink:associated_with_increased_likelihood_of   \n",
+       "...                 ...                                              ...   \n",
+       "237099     CHEBI:114785  biolink:associated_with_decreased_likelihood_of   \n",
+       "237100  UNII:52CMI0WC3Y  biolink:associated_with_decreased_likelihood_of   \n",
+       "237101     CHEBI:135738  biolink:associated_with_decreased_likelihood_of   \n",
+       "237102    MONDO:0004967  biolink:associated_with_decreased_likelihood_of   \n",
+       "237103       HP:0000360  biolink:associated_with_increased_likelihood_of   \n",
+       "\n",
+       "            object    relation                     provided_by provided_date  \\\n",
+       "0       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "1       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "2       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "3       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "4       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "...            ...         ...                             ...           ...   \n",
+       "237099  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237100  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237101  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237102  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237103  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "\n",
+       "         predicate_category           classifier   auc_roc   p_value  ...  \\\n",
+       "0       biolink:Association  Logistic Regression  0.840132  0.000000  ...   \n",
+       "1       biolink:Association  Logistic Regression  0.840132  0.999872  ...   \n",
+       "2       biolink:Association  Logistic Regression  0.840132  0.936767  ...   \n",
+       "3       biolink:Association  Logistic Regression  0.840132  0.998563  ...   \n",
+       "4       biolink:Association  Logistic Regression  0.840132  0.093959  ...   \n",
+       "...                     ...                  ...       ...       ...  ...   \n",
+       "237099  biolink:Association  Logistic Regression  0.959791  0.998503  ...   \n",
+       "237100  biolink:Association  Logistic Regression  0.959791  0.999719  ...   \n",
+       "237101  biolink:Association  Logistic Regression  0.959791  0.998357  ...   \n",
+       "237102  biolink:Association  Logistic Regression  0.959791  0.997631  ...   \n",
+       "237103  biolink:Association  Logistic Regression  0.959791  0.000000  ...   \n",
+       "\n",
+       "        feature_coefficient  num_patients_with_condition  \\\n",
+       "0                  8.796399                        99669   \n",
+       "1                  8.585212                        99938   \n",
+       "2                  4.558177                        99770   \n",
+       "3                  4.357522                        99910   \n",
+       "4                  3.926064                       100242   \n",
+       "...                     ...                          ...   \n",
+       "237099            -5.087542                         1007   \n",
+       "237100            -5.066445                         1016   \n",
+       "237101            -5.079640                          978   \n",
+       "237102            -4.683547                         1006   \n",
+       "237103             7.553003                          997   \n",
+       "\n",
+       "        num_patients_without_condition                nodes_frozenset  \\\n",
+       "0                              9999902       (HP:0000360, HP:0008629)   \n",
+       "1                             10000835    (HP:0000360, MONDO:0010643)   \n",
+       "2                             10000939  (UNII:25ADE2236L, HP:0000360)   \n",
+       "3                              9998659  (HP:0000360, UNII:K16AIQ8CTM)   \n",
+       "4                              9998750    (MONDO:0007972, HP:0000360)   \n",
+       "...                                ...                            ...   \n",
+       "237099                        10007299     (HP:0008629, CHEBI:114785)   \n",
+       "237100                         9996273  (UNII:52CMI0WC3Y, HP:0008629)   \n",
+       "237101                         9998030     (HP:0008629, CHEBI:135738)   \n",
+       "237102                        10001385    (HP:0008629, MONDO:0004967)   \n",
+       "237103                         9997731       (HP:0000360, HP:0008629)   \n",
+       "\n",
+       "             subject_id                            subject_name  \\\n",
+       "0            HP:0008629                      Pulsatile tinnitus   \n",
+       "1         MONDO:0010643                acute leukemia (disease)   \n",
+       "2       UNII:25ADE2236L                                thrombin   \n",
+       "3       UNII:K16AIQ8CTM                              pertuzumab   \n",
+       "4         MONDO:0007972                         Meniere disease   \n",
+       "...                 ...                                     ...   \n",
+       "237099     CHEBI:114785                               erlotinib   \n",
+       "237100  UNII:52CMI0WC3Y                            atezolizumab   \n",
+       "237101     CHEBI:135738                             clevidipine   \n",
+       "237102    MONDO:0004967  acute lymphoblastic leukemia (disease)   \n",
+       "237103       HP:0000360                                Tinnitus   \n",
+       "\n",
+       "                 subject_category   object_id         object_name  \\\n",
+       "0       biolink:PhenotypicFeature  HP:0000360            Tinnitus   \n",
+       "1                 biolink:Disease  HP:0000360            Tinnitus   \n",
+       "2          biolink:ChemicalEntity  HP:0000360            Tinnitus   \n",
+       "3          biolink:ChemicalEntity  HP:0000360            Tinnitus   \n",
+       "4                 biolink:Disease  HP:0000360            Tinnitus   \n",
+       "...                           ...         ...                 ...   \n",
+       "237099     biolink:ChemicalEntity  HP:0008629  Pulsatile tinnitus   \n",
+       "237100     biolink:ChemicalEntity  HP:0008629  Pulsatile tinnitus   \n",
+       "237101     biolink:ChemicalEntity  HP:0008629  Pulsatile tinnitus   \n",
+       "237102            biolink:Disease  HP:0008629  Pulsatile tinnitus   \n",
+       "237103  biolink:PhenotypicFeature  HP:0008629  Pulsatile tinnitus   \n",
+       "\n",
+       "                  object_category  \n",
+       "0       biolink:PhenotypicFeature  \n",
+       "1       biolink:PhenotypicFeature  \n",
+       "2       biolink:PhenotypicFeature  \n",
+       "3       biolink:PhenotypicFeature  \n",
+       "4       biolink:PhenotypicFeature  \n",
+       "...                           ...  \n",
+       "237099  biolink:PhenotypicFeature  \n",
+       "237100  biolink:PhenotypicFeature  \n",
+       "237101  biolink:PhenotypicFeature  \n",
+       "237102  biolink:PhenotypicFeature  \n",
+       "237103  biolink:PhenotypicFeature  \n",
+       "\n",
+       "[237104 rows x 21 columns]"
+      ]
+     },
+     "execution_count": 17,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# ENSURE THERE ARE UNIQUE RECORDS/ROWS\n",
+    "kg = kg.drop_duplicates(['subject', 'object', 'auc_roc', 'p_value', 'feature_coefficient'], keep='first')\n",
+    "kg"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 18,
+   "id": "08820f13",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "292\n",
+      "0\n"
+     ]
+    }
+   ],
+   "source": [
+    "kg_NONE_subjects = kg[kg[\"subject\"].str.contains(\"NONE\")==True]\n",
+    "print(len(kg_NONE_subjects))\n",
+    "kg_NONE_objects = kg[kg[\"object\"].str.contains(\"NONE\")==True]\n",
+    "print(len(kg_NONE_objects))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "id": "cc1c65fd",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>subject</th>\n",
+       "      <th>predicate</th>\n",
+       "      <th>object</th>\n",
+       "      <th>relation</th>\n",
+       "      <th>provided_by</th>\n",
+       "      <th>provided_date</th>\n",
+       "      <th>predicate_category</th>\n",
+       "      <th>classifier</th>\n",
+       "      <th>auc_roc</th>\n",
+       "      <th>p_value</th>\n",
+       "      <th>...</th>\n",
+       "      <th>feature_coefficient</th>\n",
+       "      <th>num_patients_with_condition</th>\n",
+       "      <th>num_patients_without_condition</th>\n",
+       "      <th>nodes_frozenset</th>\n",
+       "      <th>subject_id</th>\n",
+       "      <th>subject_name</th>\n",
+       "      <th>subject_category</th>\n",
+       "      <th>object_id</th>\n",
+       "      <th>object_name</th>\n",
+       "      <th>object_category</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>...</td>\n",
+       "      <td>8.796399</td>\n",
+       "      <td>99669</td>\n",
+       "      <td>9999902</td>\n",
+       "      <td>(HP:0000360, HP:0008629)</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>MONDO:0010643</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.999872</td>\n",
+       "      <td>...</td>\n",
+       "      <td>8.585212</td>\n",
+       "      <td>99938</td>\n",
+       "      <td>10000835</td>\n",
+       "      <td>(HP:0000360, MONDO:0010643)</td>\n",
+       "      <td>MONDO:0010643</td>\n",
+       "      <td>acute leukemia (disease)</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>UNII:25ADE2236L</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.936767</td>\n",
+       "      <td>...</td>\n",
+       "      <td>4.558177</td>\n",
+       "      <td>99770</td>\n",
+       "      <td>10000939</td>\n",
+       "      <td>(UNII:25ADE2236L, HP:0000360)</td>\n",
+       "      <td>UNII:25ADE2236L</td>\n",
+       "      <td>thrombin</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>UNII:K16AIQ8CTM</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.998563</td>\n",
+       "      <td>...</td>\n",
+       "      <td>4.357522</td>\n",
+       "      <td>99910</td>\n",
+       "      <td>9998659</td>\n",
+       "      <td>(HP:0000360, UNII:K16AIQ8CTM)</td>\n",
+       "      <td>UNII:K16AIQ8CTM</td>\n",
+       "      <td>pertuzumab</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>MONDO:0007972</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.093959</td>\n",
+       "      <td>...</td>\n",
+       "      <td>3.926064</td>\n",
+       "      <td>100242</td>\n",
+       "      <td>9998750</td>\n",
+       "      <td>(MONDO:0007972, HP:0000360)</td>\n",
+       "      <td>MONDO:0007972</td>\n",
+       "      <td>Meniere disease</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237099</th>\n",
+       "      <td>CHEBI:114785</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.998503</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-5.087542</td>\n",
+       "      <td>1007</td>\n",
+       "      <td>10007299</td>\n",
+       "      <td>(HP:0008629, CHEBI:114785)</td>\n",
+       "      <td>CHEBI:114785</td>\n",
+       "      <td>erlotinib</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237100</th>\n",
+       "      <td>UNII:52CMI0WC3Y</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.999719</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-5.066445</td>\n",
+       "      <td>1016</td>\n",
+       "      <td>9996273</td>\n",
+       "      <td>(UNII:52CMI0WC3Y, HP:0008629)</td>\n",
+       "      <td>UNII:52CMI0WC3Y</td>\n",
+       "      <td>atezolizumab</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237101</th>\n",
+       "      <td>CHEBI:135738</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.998357</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-5.079640</td>\n",
+       "      <td>978</td>\n",
+       "      <td>9998030</td>\n",
+       "      <td>(HP:0008629, CHEBI:135738)</td>\n",
+       "      <td>CHEBI:135738</td>\n",
+       "      <td>clevidipine</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237102</th>\n",
+       "      <td>MONDO:0004967</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.997631</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-4.683547</td>\n",
+       "      <td>1006</td>\n",
+       "      <td>10001385</td>\n",
+       "      <td>(HP:0008629, MONDO:0004967)</td>\n",
+       "      <td>MONDO:0004967</td>\n",
+       "      <td>acute lymphoblastic leukemia (disease)</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237103</th>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>...</td>\n",
+       "      <td>7.553003</td>\n",
+       "      <td>997</td>\n",
+       "      <td>9997731</td>\n",
+       "      <td>(HP:0000360, HP:0008629)</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>236812 rows × 21 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                subject                                        predicate  \\\n",
+       "0            HP:0008629  biolink:associated_with_increased_likelihood_of   \n",
+       "1         MONDO:0010643  biolink:associated_with_increased_likelihood_of   \n",
+       "2       UNII:25ADE2236L  biolink:associated_with_increased_likelihood_of   \n",
+       "3       UNII:K16AIQ8CTM  biolink:associated_with_increased_likelihood_of   \n",
+       "4         MONDO:0007972  biolink:associated_with_increased_likelihood_of   \n",
+       "...                 ...                                              ...   \n",
+       "237099     CHEBI:114785  biolink:associated_with_decreased_likelihood_of   \n",
+       "237100  UNII:52CMI0WC3Y  biolink:associated_with_decreased_likelihood_of   \n",
+       "237101     CHEBI:135738  biolink:associated_with_decreased_likelihood_of   \n",
+       "237102    MONDO:0004967  biolink:associated_with_decreased_likelihood_of   \n",
+       "237103       HP:0000360  biolink:associated_with_increased_likelihood_of   \n",
+       "\n",
+       "            object    relation                     provided_by provided_date  \\\n",
+       "0       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "1       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "2       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "3       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "4       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "...            ...         ...                             ...           ...   \n",
+       "237099  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237100  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237101  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237102  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237103  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "\n",
+       "         predicate_category           classifier   auc_roc   p_value  ...  \\\n",
+       "0       biolink:Association  Logistic Regression  0.840132  0.000000  ...   \n",
+       "1       biolink:Association  Logistic Regression  0.840132  0.999872  ...   \n",
+       "2       biolink:Association  Logistic Regression  0.840132  0.936767  ...   \n",
+       "3       biolink:Association  Logistic Regression  0.840132  0.998563  ...   \n",
+       "4       biolink:Association  Logistic Regression  0.840132  0.093959  ...   \n",
+       "...                     ...                  ...       ...       ...  ...   \n",
+       "237099  biolink:Association  Logistic Regression  0.959791  0.998503  ...   \n",
+       "237100  biolink:Association  Logistic Regression  0.959791  0.999719  ...   \n",
+       "237101  biolink:Association  Logistic Regression  0.959791  0.998357  ...   \n",
+       "237102  biolink:Association  Logistic Regression  0.959791  0.997631  ...   \n",
+       "237103  biolink:Association  Logistic Regression  0.959791  0.000000  ...   \n",
+       "\n",
+       "        feature_coefficient  num_patients_with_condition  \\\n",
+       "0                  8.796399                        99669   \n",
+       "1                  8.585212                        99938   \n",
+       "2                  4.558177                        99770   \n",
+       "3                  4.357522                        99910   \n",
+       "4                  3.926064                       100242   \n",
+       "...                     ...                          ...   \n",
+       "237099            -5.087542                         1007   \n",
+       "237100            -5.066445                         1016   \n",
+       "237101            -5.079640                          978   \n",
+       "237102            -4.683547                         1006   \n",
+       "237103             7.553003                          997   \n",
+       "\n",
+       "        num_patients_without_condition                nodes_frozenset  \\\n",
+       "0                              9999902       (HP:0000360, HP:0008629)   \n",
+       "1                             10000835    (HP:0000360, MONDO:0010643)   \n",
+       "2                             10000939  (UNII:25ADE2236L, HP:0000360)   \n",
+       "3                              9998659  (HP:0000360, UNII:K16AIQ8CTM)   \n",
+       "4                              9998750    (MONDO:0007972, HP:0000360)   \n",
+       "...                                ...                            ...   \n",
+       "237099                        10007299     (HP:0008629, CHEBI:114785)   \n",
+       "237100                         9996273  (UNII:52CMI0WC3Y, HP:0008629)   \n",
+       "237101                         9998030     (HP:0008629, CHEBI:135738)   \n",
+       "237102                        10001385    (HP:0008629, MONDO:0004967)   \n",
+       "237103                         9997731       (HP:0000360, HP:0008629)   \n",
+       "\n",
+       "             subject_id                            subject_name  \\\n",
+       "0            HP:0008629                      Pulsatile tinnitus   \n",
+       "1         MONDO:0010643                acute leukemia (disease)   \n",
+       "2       UNII:25ADE2236L                                thrombin   \n",
+       "3       UNII:K16AIQ8CTM                              pertuzumab   \n",
+       "4         MONDO:0007972                         Meniere disease   \n",
+       "...                 ...                                     ...   \n",
+       "237099     CHEBI:114785                               erlotinib   \n",
+       "237100  UNII:52CMI0WC3Y                            atezolizumab   \n",
+       "237101     CHEBI:135738                             clevidipine   \n",
+       "237102    MONDO:0004967  acute lymphoblastic leukemia (disease)   \n",
+       "237103       HP:0000360                                Tinnitus   \n",
+       "\n",
+       "                 subject_category   object_id         object_name  \\\n",
+       "0       biolink:PhenotypicFeature  HP:0000360            Tinnitus   \n",
+       "1                 biolink:Disease  HP:0000360            Tinnitus   \n",
+       "2          biolink:ChemicalEntity  HP:0000360            Tinnitus   \n",
+       "3          biolink:ChemicalEntity  HP:0000360            Tinnitus   \n",
+       "4                 biolink:Disease  HP:0000360            Tinnitus   \n",
+       "...                           ...         ...                 ...   \n",
+       "237099     biolink:ChemicalEntity  HP:0008629  Pulsatile tinnitus   \n",
+       "237100     biolink:ChemicalEntity  HP:0008629  Pulsatile tinnitus   \n",
+       "237101     biolink:ChemicalEntity  HP:0008629  Pulsatile tinnitus   \n",
+       "237102            biolink:Disease  HP:0008629  Pulsatile tinnitus   \n",
+       "237103  biolink:PhenotypicFeature  HP:0008629  Pulsatile tinnitus   \n",
+       "\n",
+       "                  object_category  \n",
+       "0       biolink:PhenotypicFeature  \n",
+       "1       biolink:PhenotypicFeature  \n",
+       "2       biolink:PhenotypicFeature  \n",
+       "3       biolink:PhenotypicFeature  \n",
+       "4       biolink:PhenotypicFeature  \n",
+       "...                           ...  \n",
+       "237099  biolink:PhenotypicFeature  \n",
+       "237100  biolink:PhenotypicFeature  \n",
+       "237101  biolink:PhenotypicFeature  \n",
+       "237102  biolink:PhenotypicFeature  \n",
+       "237103  biolink:PhenotypicFeature  \n",
+       "\n",
+       "[236812 rows x 21 columns]"
+      ]
+     },
+     "execution_count": 19,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# for some reason, some subject or object ids are empty (they actually literally contain the string \"NONE\")\n",
+    "kg = kg.dropna(axis=0, subset=['subject'])\n",
+    "kg = kg.dropna(axis=0, subset=['object'])\n",
+    "kg = kg[~kg[\"subject\"].str.contains(\"NONE\")==True]\n",
+    "kg = kg[~kg[\"object\"].str.contains(\"NONE\")==True]\n",
+    "kg"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 20,
+   "id": "c14460e0",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>subject</th>\n",
+       "      <th>predicate</th>\n",
+       "      <th>object</th>\n",
+       "      <th>relation</th>\n",
+       "      <th>provided_by</th>\n",
+       "      <th>provided_date</th>\n",
+       "      <th>predicate_category</th>\n",
+       "      <th>classifier</th>\n",
+       "      <th>auc_roc</th>\n",
+       "      <th>p_value</th>\n",
+       "      <th>...</th>\n",
+       "      <th>feature_coefficient</th>\n",
+       "      <th>num_patients_with_condition</th>\n",
+       "      <th>num_patients_without_condition</th>\n",
+       "      <th>nodes_frozenset</th>\n",
+       "      <th>subject_id</th>\n",
+       "      <th>subject_name</th>\n",
+       "      <th>subject_category</th>\n",
+       "      <th>object_id</th>\n",
+       "      <th>object_name</th>\n",
+       "      <th>object_category</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>...</td>\n",
+       "      <td>8.796399</td>\n",
+       "      <td>99669</td>\n",
+       "      <td>9999902</td>\n",
+       "      <td>(HP:0000360, HP:0008629)</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>MONDO:0010643</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.999872</td>\n",
+       "      <td>...</td>\n",
+       "      <td>8.585212</td>\n",
+       "      <td>99938</td>\n",
+       "      <td>10000835</td>\n",
+       "      <td>(HP:0000360, MONDO:0010643)</td>\n",
+       "      <td>MONDO:0010643</td>\n",
+       "      <td>acute leukemia (disease)</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>UNII:25ADE2236L</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.936767</td>\n",
+       "      <td>...</td>\n",
+       "      <td>4.558177</td>\n",
+       "      <td>99770</td>\n",
+       "      <td>10000939</td>\n",
+       "      <td>(UNII:25ADE2236L, HP:0000360)</td>\n",
+       "      <td>UNII:25ADE2236L</td>\n",
+       "      <td>thrombin</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>UNII:K16AIQ8CTM</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.998563</td>\n",
+       "      <td>...</td>\n",
+       "      <td>4.357522</td>\n",
+       "      <td>99910</td>\n",
+       "      <td>9998659</td>\n",
+       "      <td>(HP:0000360, UNII:K16AIQ8CTM)</td>\n",
+       "      <td>UNII:K16AIQ8CTM</td>\n",
+       "      <td>pertuzumab</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>MONDO:0007972</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.093959</td>\n",
+       "      <td>...</td>\n",
+       "      <td>3.926064</td>\n",
+       "      <td>100242</td>\n",
+       "      <td>9998750</td>\n",
+       "      <td>(MONDO:0007972, HP:0000360)</td>\n",
+       "      <td>MONDO:0007972</td>\n",
+       "      <td>Meniere disease</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237099</th>\n",
+       "      <td>CHEBI:114785</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.998503</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-5.087542</td>\n",
+       "      <td>1007</td>\n",
+       "      <td>10007299</td>\n",
+       "      <td>(HP:0008629, CHEBI:114785)</td>\n",
+       "      <td>CHEBI:114785</td>\n",
+       "      <td>erlotinib</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237100</th>\n",
+       "      <td>UNII:52CMI0WC3Y</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.999719</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-5.066445</td>\n",
+       "      <td>1016</td>\n",
+       "      <td>9996273</td>\n",
+       "      <td>(UNII:52CMI0WC3Y, HP:0008629)</td>\n",
+       "      <td>UNII:52CMI0WC3Y</td>\n",
+       "      <td>atezolizumab</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237101</th>\n",
+       "      <td>CHEBI:135738</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.998357</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-5.079640</td>\n",
+       "      <td>978</td>\n",
+       "      <td>9998030</td>\n",
+       "      <td>(HP:0008629, CHEBI:135738)</td>\n",
+       "      <td>CHEBI:135738</td>\n",
+       "      <td>clevidipine</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237102</th>\n",
+       "      <td>MONDO:0004967</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.997631</td>\n",
+       "      <td>...</td>\n",
+       "      <td>-4.683547</td>\n",
+       "      <td>1006</td>\n",
+       "      <td>10001385</td>\n",
+       "      <td>(HP:0008629, MONDO:0004967)</td>\n",
+       "      <td>MONDO:0004967</td>\n",
+       "      <td>acute lymphoblastic leukemia (disease)</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237103</th>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>...</td>\n",
+       "      <td>7.553003</td>\n",
+       "      <td>997</td>\n",
+       "      <td>9997731</td>\n",
+       "      <td>(HP:0000360, HP:0008629)</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>236812 rows × 21 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                subject                                        predicate  \\\n",
+       "0            HP:0008629  biolink:associated_with_increased_likelihood_of   \n",
+       "1         MONDO:0010643  biolink:associated_with_increased_likelihood_of   \n",
+       "2       UNII:25ADE2236L  biolink:associated_with_increased_likelihood_of   \n",
+       "3       UNII:K16AIQ8CTM  biolink:associated_with_increased_likelihood_of   \n",
+       "4         MONDO:0007972  biolink:associated_with_increased_likelihood_of   \n",
+       "...                 ...                                              ...   \n",
+       "237099     CHEBI:114785  biolink:associated_with_decreased_likelihood_of   \n",
+       "237100  UNII:52CMI0WC3Y  biolink:associated_with_decreased_likelihood_of   \n",
+       "237101     CHEBI:135738  biolink:associated_with_decreased_likelihood_of   \n",
+       "237102    MONDO:0004967  biolink:associated_with_decreased_likelihood_of   \n",
+       "237103       HP:0000360  biolink:associated_with_increased_likelihood_of   \n",
+       "\n",
+       "            object    relation                     provided_by provided_date  \\\n",
+       "0       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "1       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "2       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "3       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "4       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "...            ...         ...                             ...           ...   \n",
+       "237099  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237100  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237101  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237102  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237103  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "\n",
+       "         predicate_category           classifier   auc_roc   p_value  ...  \\\n",
+       "0       biolink:Association  Logistic Regression  0.840132  0.000000  ...   \n",
+       "1       biolink:Association  Logistic Regression  0.840132  0.999872  ...   \n",
+       "2       biolink:Association  Logistic Regression  0.840132  0.936767  ...   \n",
+       "3       biolink:Association  Logistic Regression  0.840132  0.998563  ...   \n",
+       "4       biolink:Association  Logistic Regression  0.840132  0.093959  ...   \n",
+       "...                     ...                  ...       ...       ...  ...   \n",
+       "237099  biolink:Association  Logistic Regression  0.959791  0.998503  ...   \n",
+       "237100  biolink:Association  Logistic Regression  0.959791  0.999719  ...   \n",
+       "237101  biolink:Association  Logistic Regression  0.959791  0.998357  ...   \n",
+       "237102  biolink:Association  Logistic Regression  0.959791  0.997631  ...   \n",
+       "237103  biolink:Association  Logistic Regression  0.959791  0.000000  ...   \n",
+       "\n",
+       "        feature_coefficient  num_patients_with_condition  \\\n",
+       "0                  8.796399                        99669   \n",
+       "1                  8.585212                        99938   \n",
+       "2                  4.558177                        99770   \n",
+       "3                  4.357522                        99910   \n",
+       "4                  3.926064                       100242   \n",
+       "...                     ...                          ...   \n",
+       "237099            -5.087542                         1007   \n",
+       "237100            -5.066445                         1016   \n",
+       "237101            -5.079640                          978   \n",
+       "237102            -4.683547                         1006   \n",
+       "237103             7.553003                          997   \n",
+       "\n",
+       "        num_patients_without_condition                nodes_frozenset  \\\n",
+       "0                              9999902       (HP:0000360, HP:0008629)   \n",
+       "1                             10000835    (HP:0000360, MONDO:0010643)   \n",
+       "2                             10000939  (UNII:25ADE2236L, HP:0000360)   \n",
+       "3                              9998659  (HP:0000360, UNII:K16AIQ8CTM)   \n",
+       "4                              9998750    (MONDO:0007972, HP:0000360)   \n",
+       "...                                ...                            ...   \n",
+       "237099                        10007299     (HP:0008629, CHEBI:114785)   \n",
+       "237100                         9996273  (UNII:52CMI0WC3Y, HP:0008629)   \n",
+       "237101                         9998030     (HP:0008629, CHEBI:135738)   \n",
+       "237102                        10001385    (HP:0008629, MONDO:0004967)   \n",
+       "237103                         9997731       (HP:0000360, HP:0008629)   \n",
+       "\n",
+       "             subject_id                            subject_name  \\\n",
+       "0            HP:0008629                      Pulsatile tinnitus   \n",
+       "1         MONDO:0010643                acute leukemia (disease)   \n",
+       "2       UNII:25ADE2236L                                thrombin   \n",
+       "3       UNII:K16AIQ8CTM                              pertuzumab   \n",
+       "4         MONDO:0007972                         Meniere disease   \n",
+       "...                 ...                                     ...   \n",
+       "237099     CHEBI:114785                               erlotinib   \n",
+       "237100  UNII:52CMI0WC3Y                            atezolizumab   \n",
+       "237101     CHEBI:135738                             clevidipine   \n",
+       "237102    MONDO:0004967  acute lymphoblastic leukemia (disease)   \n",
+       "237103       HP:0000360                                Tinnitus   \n",
+       "\n",
+       "                 subject_category   object_id         object_name  \\\n",
+       "0       biolink:PhenotypicFeature  HP:0000360            Tinnitus   \n",
+       "1                 biolink:Disease  HP:0000360            Tinnitus   \n",
+       "2          biolink:ChemicalEntity  HP:0000360            Tinnitus   \n",
+       "3          biolink:ChemicalEntity  HP:0000360            Tinnitus   \n",
+       "4                 biolink:Disease  HP:0000360            Tinnitus   \n",
+       "...                           ...         ...                 ...   \n",
+       "237099     biolink:ChemicalEntity  HP:0008629  Pulsatile tinnitus   \n",
+       "237100     biolink:ChemicalEntity  HP:0008629  Pulsatile tinnitus   \n",
+       "237101     biolink:ChemicalEntity  HP:0008629  Pulsatile tinnitus   \n",
+       "237102            biolink:Disease  HP:0008629  Pulsatile tinnitus   \n",
+       "237103  biolink:PhenotypicFeature  HP:0008629  Pulsatile tinnitus   \n",
+       "\n",
+       "                  object_category  \n",
+       "0       biolink:PhenotypicFeature  \n",
+       "1       biolink:PhenotypicFeature  \n",
+       "2       biolink:PhenotypicFeature  \n",
+       "3       biolink:PhenotypicFeature  \n",
+       "4       biolink:PhenotypicFeature  \n",
+       "...                           ...  \n",
+       "237099  biolink:PhenotypicFeature  \n",
+       "237100  biolink:PhenotypicFeature  \n",
+       "237101  biolink:PhenotypicFeature  \n",
+       "237102  biolink:PhenotypicFeature  \n",
+       "237103  biolink:PhenotypicFeature  \n",
+       "\n",
+       "[236812 rows x 21 columns]"
+      ]
+     },
+     "execution_count": 20,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# for some reason, some subject or object ids are empty (they actually literally contain the string \"NONE\")\n",
+    "# get rid of these rows\n",
+    "kg = kg[~kg[\"subject\"].str.contains(\"NONE\")==True]  # subject and object are all CURIEs, not names\n",
+    "kg = kg[~kg[\"object\"].str.contains(\"NONE\")==True]\n",
+    "kg = kg[~kg[\"subject\"].str.contains(\"none\")==True]\n",
+    "kg = kg[~kg[\"object\"].str.contains(\"none\")==True]\n",
+    "kg = kg[~kg[\"subject\"].str.contains(\"None\")==True]\n",
+    "kg = kg[~kg[\"object\"].str.contains(\"None\")==True]\n",
+    "kg"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "5201401a",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# OPTIONAL THRESHOLD\n",
+    "# kg_pval_subsetted = kg[kg[\"p_value\"] < 0.2]\n",
+    "# kg_pval_subsetted"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 334,
+   "id": "e809e4a0",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>subject_category_postfix</th>\n",
+       "      <th>subject_prefix</th>\n",
+       "      <th>increased_or_decreased</th>\n",
+       "      <th>object_category_postfix</th>\n",
+       "      <th>object_prefix</th>\n",
+       "      <th>Count</th>\n",
+       "      <th>object_category_postfix_abbv</th>\n",
+       "      <th>subject_category_postfix_abbv</th>\n",
+       "      <th>joined_to_match_xbte</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>CHEBI</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>34917</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemCHEBI_decreased_DiseaseMONDO</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>CHEBI</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>212</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemCHEBI_decreased_DiseaseNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>CHEBI</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>2917</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemCHEBI_decreased_DiseaseSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>CHEBI</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>24505</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemCHEBI_decreased_PhenoHP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>CHEBI</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>2254</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemCHEBI_decreased_PhenoNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>5</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>CHEBI</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>244</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemCHEBI_decreased_PhenoSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>6</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>CHEBI</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>222</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemCHEBI_decreased_ProcedureNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>7</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>CHEBI</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>18248</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemCHEBI_increased_DiseaseMONDO</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>8</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>CHEBI</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>131</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemCHEBI_increased_DiseaseNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>9</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>CHEBI</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>1542</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemCHEBI_increased_DiseaseSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>10</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>CHEBI</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>13225</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemCHEBI_increased_PhenoHP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>11</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>CHEBI</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>1176</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemCHEBI_increased_PhenoNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>12</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>CHEBI</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>99</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemCHEBI_increased_PhenoSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>13</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>CHEBI</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>121</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemCHEBI_increased_ProcedureNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>14</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>UNII</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>5902</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemUNII_decreased_DiseaseMONDO</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>15</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>UNII</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>44</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemUNII_decreased_DiseaseNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>16</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>UNII</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>505</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemUNII_decreased_DiseaseSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>17</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>UNII</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>3727</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemUNII_decreased_PhenoHP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>18</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>UNII</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>381</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemUNII_decreased_PhenoNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>19</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>UNII</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>39</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemUNII_decreased_PhenoSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>20</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>UNII</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>43</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemUNII_decreased_ProcedureNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>21</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>UNII</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>2158</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemUNII_increased_DiseaseMONDO</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>22</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>UNII</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>8</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemUNII_increased_DiseaseNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>23</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>UNII</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>171</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemUNII_increased_DiseaseSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>24</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>UNII</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>1993</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemUNII_increased_PhenoHP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>25</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>UNII</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>139</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemUNII_increased_PhenoNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>26</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>UNII</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>13</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemUNII_increased_PhenoSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>27</th>\n",
+       "      <td>ChemicalEntity</td>\n",
+       "      <td>UNII</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>9</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>Chem</td>\n",
+       "      <td>ChemUNII_increased_ProcedureNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>28</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>13640</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseMONDO_decreased_DiseaseMONDO</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>29</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>86</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseMONDO_decreased_DiseaseNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>30</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>1237</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseMONDO_decreased_DiseaseSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>31</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>10101</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseMONDO_decreased_PhenoHP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>32</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>1293</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseMONDO_decreased_PhenoNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>33</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>88</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseMONDO_decreased_PhenoSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>34</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>113</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseMONDO_decreased_ProcedureNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>35</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>14415</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseMONDO_increased_DiseaseMONDO</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>36</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>96</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseMONDO_increased_DiseaseNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>37</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>1129</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseMONDO_increased_DiseaseSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>38</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>9919</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseMONDO_increased_PhenoHP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>39</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>527</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseMONDO_increased_PhenoNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>40</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>94</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseMONDO_increased_PhenoSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>41</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>69</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseMONDO_increased_ProcedureNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>42</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>73</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseNCIT_decreased_DiseaseMONDO</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>43</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>7</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseNCIT_decreased_DiseaseSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>44</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>44</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseNCIT_decreased_PhenoHP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>45</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>9</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseNCIT_decreased_PhenoNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>46</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseNCIT_decreased_ProcedureNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>47</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>82</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseNCIT_increased_DiseaseMONDO</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>48</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>6</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseNCIT_increased_DiseaseSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>49</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>66</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseNCIT_increased_PhenoHP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>50</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseNCIT_increased_PhenoNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>51</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseNCIT_increased_PhenoSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>52</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>1318</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseSNOMEDCT_decreased_DiseaseMONDO</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>53</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>10</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseSNOMEDCT_decreased_DiseaseNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>54</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>101</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseSNOMEDCT_decreased_DiseaseSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>55</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>957</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseSNOMEDCT_decreased_PhenoHP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>56</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>116</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseSNOMEDCT_decreased_PhenoNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>57</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>8</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseSNOMEDCT_decreased_PhenoSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>58</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>13</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseSNOMEDCT_decreased_ProcedureNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>59</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>1007</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseSNOMEDCT_increased_DiseaseMONDO</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>60</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>5</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseSNOMEDCT_increased_DiseaseNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>61</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>81</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseSNOMEDCT_increased_DiseaseSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>62</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>693</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseSNOMEDCT_increased_PhenoHP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>63</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>34</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseSNOMEDCT_increased_PhenoNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>64</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>7</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseSNOMEDCT_increased_PhenoSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>65</th>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>2</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>DiseaseSNOMEDCT_increased_ProcedureNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>66</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>18759</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoHP_decreased_DiseaseMONDO</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>67</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>113</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoHP_decreased_DiseaseNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>68</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>1532</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoHP_decreased_DiseaseSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>69</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>11425</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoHP_decreased_PhenoHP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>70</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>1257</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoHP_decreased_PhenoNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>71</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>121</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoHP_decreased_PhenoSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>72</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>132</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoHP_decreased_ProcedureNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>73</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>12086</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoHP_increased_DiseaseMONDO</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>74</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>86</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoHP_increased_DiseaseNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>75</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>1055</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoHP_increased_DiseaseSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>76</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>10355</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoHP_increased_PhenoHP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>77</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>733</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoHP_increased_PhenoNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>78</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>78</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoHP_increased_PhenoSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>79</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>67</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoHP_increased_ProcedureNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>80</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>1170</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoNCIT_decreased_DiseaseMONDO</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>81</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>7</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoNCIT_decreased_DiseaseNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>82</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>94</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoNCIT_decreased_DiseaseSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>83</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>705</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoNCIT_decreased_PhenoHP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>84</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>56</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoNCIT_decreased_PhenoNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>85</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>9</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoNCIT_decreased_PhenoSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>86</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>7</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoNCIT_decreased_ProcedureNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>87</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>380</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoNCIT_increased_DiseaseMONDO</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>88</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>3</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoNCIT_increased_DiseaseNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>89</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>36</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoNCIT_increased_DiseaseSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>90</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>395</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoNCIT_increased_PhenoHP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>91</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>34</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoNCIT_increased_PhenoNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>92</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoNCIT_increased_PhenoSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>93</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>3</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoNCIT_increased_ProcedureNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>94</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>307</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoSNOMEDCT_decreased_DiseaseMONDO</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>95</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoSNOMEDCT_decreased_DiseaseNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>96</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>26</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoSNOMEDCT_decreased_DiseaseSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>97</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>217</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoSNOMEDCT_decreased_PhenoHP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>98</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>25</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoSNOMEDCT_decreased_PhenoNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>99</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>2</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoSNOMEDCT_decreased_PhenoSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>100</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>2</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoSNOMEDCT_decreased_ProcedureNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>101</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>158</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoSNOMEDCT_increased_DiseaseMONDO</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>102</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>2</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoSNOMEDCT_increased_DiseaseNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>103</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>13</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoSNOMEDCT_increased_DiseaseSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>104</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>113</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoSNOMEDCT_increased_PhenoHP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>105</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>5</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoSNOMEDCT_increased_PhenoNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>106</th>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>PhenoSNOMEDCT_increased_ProcedureNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>107</th>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>721</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>ProcedureNCIT_decreased_DiseaseMONDO</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>108</th>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>4</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>ProcedureNCIT_decreased_DiseaseNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>109</th>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>58</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>ProcedureNCIT_decreased_DiseaseSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>110</th>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>502</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>ProcedureNCIT_decreased_PhenoHP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>111</th>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>56</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>ProcedureNCIT_decreased_PhenoNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>112</th>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>4</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>ProcedureNCIT_decreased_PhenoSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>113</th>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>decreased</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>3</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>ProcedureNCIT_decreased_ProcedureNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>114</th>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>209</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>ProcedureNCIT_increased_DiseaseMONDO</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>115</th>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>2</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>ProcedureNCIT_increased_DiseaseNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>116</th>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>20</td>\n",
+       "      <td>Disease</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>ProcedureNCIT_increased_DiseaseSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>117</th>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>158</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>ProcedureNCIT_increased_PhenoHP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>118</th>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>4</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>ProcedureNCIT_increased_PhenoNCIT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>119</th>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>PhenotypicFeature</td>\n",
+       "      <td>SNOMEDCT</td>\n",
+       "      <td>2</td>\n",
+       "      <td>Pheno</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>ProcedureNCIT_increased_PhenoSNOMEDCT</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>120</th>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>increased</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>NCIT</td>\n",
+       "      <td>2</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>Procedure</td>\n",
+       "      <td>ProcedureNCIT_increased_ProcedureNCIT</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "    subject_category_postfix subject_prefix increased_or_decreased  \\\n",
+       "0             ChemicalEntity          CHEBI              decreased   \n",
+       "1             ChemicalEntity          CHEBI              decreased   \n",
+       "2             ChemicalEntity          CHEBI              decreased   \n",
+       "3             ChemicalEntity          CHEBI              decreased   \n",
+       "4             ChemicalEntity          CHEBI              decreased   \n",
+       "5             ChemicalEntity          CHEBI              decreased   \n",
+       "6             ChemicalEntity          CHEBI              decreased   \n",
+       "7             ChemicalEntity          CHEBI              increased   \n",
+       "8             ChemicalEntity          CHEBI              increased   \n",
+       "9             ChemicalEntity          CHEBI              increased   \n",
+       "10            ChemicalEntity          CHEBI              increased   \n",
+       "11            ChemicalEntity          CHEBI              increased   \n",
+       "12            ChemicalEntity          CHEBI              increased   \n",
+       "13            ChemicalEntity          CHEBI              increased   \n",
+       "14            ChemicalEntity           UNII              decreased   \n",
+       "15            ChemicalEntity           UNII              decreased   \n",
+       "16            ChemicalEntity           UNII              decreased   \n",
+       "17            ChemicalEntity           UNII              decreased   \n",
+       "18            ChemicalEntity           UNII              decreased   \n",
+       "19            ChemicalEntity           UNII              decreased   \n",
+       "20            ChemicalEntity           UNII              decreased   \n",
+       "21            ChemicalEntity           UNII              increased   \n",
+       "22            ChemicalEntity           UNII              increased   \n",
+       "23            ChemicalEntity           UNII              increased   \n",
+       "24            ChemicalEntity           UNII              increased   \n",
+       "25            ChemicalEntity           UNII              increased   \n",
+       "26            ChemicalEntity           UNII              increased   \n",
+       "27            ChemicalEntity           UNII              increased   \n",
+       "28                   Disease          MONDO              decreased   \n",
+       "29                   Disease          MONDO              decreased   \n",
+       "30                   Disease          MONDO              decreased   \n",
+       "31                   Disease          MONDO              decreased   \n",
+       "32                   Disease          MONDO              decreased   \n",
+       "33                   Disease          MONDO              decreased   \n",
+       "34                   Disease          MONDO              decreased   \n",
+       "35                   Disease          MONDO              increased   \n",
+       "36                   Disease          MONDO              increased   \n",
+       "37                   Disease          MONDO              increased   \n",
+       "38                   Disease          MONDO              increased   \n",
+       "39                   Disease          MONDO              increased   \n",
+       "40                   Disease          MONDO              increased   \n",
+       "41                   Disease          MONDO              increased   \n",
+       "42                   Disease           NCIT              decreased   \n",
+       "43                   Disease           NCIT              decreased   \n",
+       "44                   Disease           NCIT              decreased   \n",
+       "45                   Disease           NCIT              decreased   \n",
+       "46                   Disease           NCIT              decreased   \n",
+       "47                   Disease           NCIT              increased   \n",
+       "48                   Disease           NCIT              increased   \n",
+       "49                   Disease           NCIT              increased   \n",
+       "50                   Disease           NCIT              increased   \n",
+       "51                   Disease           NCIT              increased   \n",
+       "52                   Disease       SNOMEDCT              decreased   \n",
+       "53                   Disease       SNOMEDCT              decreased   \n",
+       "54                   Disease       SNOMEDCT              decreased   \n",
+       "55                   Disease       SNOMEDCT              decreased   \n",
+       "56                   Disease       SNOMEDCT              decreased   \n",
+       "57                   Disease       SNOMEDCT              decreased   \n",
+       "58                   Disease       SNOMEDCT              decreased   \n",
+       "59                   Disease       SNOMEDCT              increased   \n",
+       "60                   Disease       SNOMEDCT              increased   \n",
+       "61                   Disease       SNOMEDCT              increased   \n",
+       "62                   Disease       SNOMEDCT              increased   \n",
+       "63                   Disease       SNOMEDCT              increased   \n",
+       "64                   Disease       SNOMEDCT              increased   \n",
+       "65                   Disease       SNOMEDCT              increased   \n",
+       "66         PhenotypicFeature             HP              decreased   \n",
+       "67         PhenotypicFeature             HP              decreased   \n",
+       "68         PhenotypicFeature             HP              decreased   \n",
+       "69         PhenotypicFeature             HP              decreased   \n",
+       "70         PhenotypicFeature             HP              decreased   \n",
+       "71         PhenotypicFeature             HP              decreased   \n",
+       "72         PhenotypicFeature             HP              decreased   \n",
+       "73         PhenotypicFeature             HP              increased   \n",
+       "74         PhenotypicFeature             HP              increased   \n",
+       "75         PhenotypicFeature             HP              increased   \n",
+       "76         PhenotypicFeature             HP              increased   \n",
+       "77         PhenotypicFeature             HP              increased   \n",
+       "78         PhenotypicFeature             HP              increased   \n",
+       "79         PhenotypicFeature             HP              increased   \n",
+       "80         PhenotypicFeature           NCIT              decreased   \n",
+       "81         PhenotypicFeature           NCIT              decreased   \n",
+       "82         PhenotypicFeature           NCIT              decreased   \n",
+       "83         PhenotypicFeature           NCIT              decreased   \n",
+       "84         PhenotypicFeature           NCIT              decreased   \n",
+       "85         PhenotypicFeature           NCIT              decreased   \n",
+       "86         PhenotypicFeature           NCIT              decreased   \n",
+       "87         PhenotypicFeature           NCIT              increased   \n",
+       "88         PhenotypicFeature           NCIT              increased   \n",
+       "89         PhenotypicFeature           NCIT              increased   \n",
+       "90         PhenotypicFeature           NCIT              increased   \n",
+       "91         PhenotypicFeature           NCIT              increased   \n",
+       "92         PhenotypicFeature           NCIT              increased   \n",
+       "93         PhenotypicFeature           NCIT              increased   \n",
+       "94         PhenotypicFeature       SNOMEDCT              decreased   \n",
+       "95         PhenotypicFeature       SNOMEDCT              decreased   \n",
+       "96         PhenotypicFeature       SNOMEDCT              decreased   \n",
+       "97         PhenotypicFeature       SNOMEDCT              decreased   \n",
+       "98         PhenotypicFeature       SNOMEDCT              decreased   \n",
+       "99         PhenotypicFeature       SNOMEDCT              decreased   \n",
+       "100        PhenotypicFeature       SNOMEDCT              decreased   \n",
+       "101        PhenotypicFeature       SNOMEDCT              increased   \n",
+       "102        PhenotypicFeature       SNOMEDCT              increased   \n",
+       "103        PhenotypicFeature       SNOMEDCT              increased   \n",
+       "104        PhenotypicFeature       SNOMEDCT              increased   \n",
+       "105        PhenotypicFeature       SNOMEDCT              increased   \n",
+       "106        PhenotypicFeature       SNOMEDCT              increased   \n",
+       "107                Procedure           NCIT              decreased   \n",
+       "108                Procedure           NCIT              decreased   \n",
+       "109                Procedure           NCIT              decreased   \n",
+       "110                Procedure           NCIT              decreased   \n",
+       "111                Procedure           NCIT              decreased   \n",
+       "112                Procedure           NCIT              decreased   \n",
+       "113                Procedure           NCIT              decreased   \n",
+       "114                Procedure           NCIT              increased   \n",
+       "115                Procedure           NCIT              increased   \n",
+       "116                Procedure           NCIT              increased   \n",
+       "117                Procedure           NCIT              increased   \n",
+       "118                Procedure           NCIT              increased   \n",
+       "119                Procedure           NCIT              increased   \n",
+       "120                Procedure           NCIT              increased   \n",
+       "\n",
+       "    object_category_postfix object_prefix  Count object_category_postfix_abbv  \\\n",
+       "0                   Disease         MONDO  34917                      Disease   \n",
+       "1                   Disease          NCIT    212                      Disease   \n",
+       "2                   Disease      SNOMEDCT   2917                      Disease   \n",
+       "3         PhenotypicFeature            HP  24505                        Pheno   \n",
+       "4         PhenotypicFeature          NCIT   2254                        Pheno   \n",
+       "5         PhenotypicFeature      SNOMEDCT    244                        Pheno   \n",
+       "6                 Procedure          NCIT    222                    Procedure   \n",
+       "7                   Disease         MONDO  18248                      Disease   \n",
+       "8                   Disease          NCIT    131                      Disease   \n",
+       "9                   Disease      SNOMEDCT   1542                      Disease   \n",
+       "10        PhenotypicFeature            HP  13225                        Pheno   \n",
+       "11        PhenotypicFeature          NCIT   1176                        Pheno   \n",
+       "12        PhenotypicFeature      SNOMEDCT     99                        Pheno   \n",
+       "13                Procedure          NCIT    121                    Procedure   \n",
+       "14                  Disease         MONDO   5902                      Disease   \n",
+       "15                  Disease          NCIT     44                      Disease   \n",
+       "16                  Disease      SNOMEDCT    505                      Disease   \n",
+       "17        PhenotypicFeature            HP   3727                        Pheno   \n",
+       "18        PhenotypicFeature          NCIT    381                        Pheno   \n",
+       "19        PhenotypicFeature      SNOMEDCT     39                        Pheno   \n",
+       "20                Procedure          NCIT     43                    Procedure   \n",
+       "21                  Disease         MONDO   2158                      Disease   \n",
+       "22                  Disease          NCIT      8                      Disease   \n",
+       "23                  Disease      SNOMEDCT    171                      Disease   \n",
+       "24        PhenotypicFeature            HP   1993                        Pheno   \n",
+       "25        PhenotypicFeature          NCIT    139                        Pheno   \n",
+       "26        PhenotypicFeature      SNOMEDCT     13                        Pheno   \n",
+       "27                Procedure          NCIT      9                    Procedure   \n",
+       "28                  Disease         MONDO  13640                      Disease   \n",
+       "29                  Disease          NCIT     86                      Disease   \n",
+       "30                  Disease      SNOMEDCT   1237                      Disease   \n",
+       "31        PhenotypicFeature            HP  10101                        Pheno   \n",
+       "32        PhenotypicFeature          NCIT   1293                        Pheno   \n",
+       "33        PhenotypicFeature      SNOMEDCT     88                        Pheno   \n",
+       "34                Procedure          NCIT    113                    Procedure   \n",
+       "35                  Disease         MONDO  14415                      Disease   \n",
+       "36                  Disease          NCIT     96                      Disease   \n",
+       "37                  Disease      SNOMEDCT   1129                      Disease   \n",
+       "38        PhenotypicFeature            HP   9919                        Pheno   \n",
+       "39        PhenotypicFeature          NCIT    527                        Pheno   \n",
+       "40        PhenotypicFeature      SNOMEDCT     94                        Pheno   \n",
+       "41                Procedure          NCIT     69                    Procedure   \n",
+       "42                  Disease         MONDO     73                      Disease   \n",
+       "43                  Disease      SNOMEDCT      7                      Disease   \n",
+       "44        PhenotypicFeature            HP     44                        Pheno   \n",
+       "45        PhenotypicFeature          NCIT      9                        Pheno   \n",
+       "46                Procedure          NCIT      1                    Procedure   \n",
+       "47                  Disease         MONDO     82                      Disease   \n",
+       "48                  Disease      SNOMEDCT      6                      Disease   \n",
+       "49        PhenotypicFeature            HP     66                        Pheno   \n",
+       "50        PhenotypicFeature          NCIT      1                        Pheno   \n",
+       "51        PhenotypicFeature      SNOMEDCT      1                        Pheno   \n",
+       "52                  Disease         MONDO   1318                      Disease   \n",
+       "53                  Disease          NCIT     10                      Disease   \n",
+       "54                  Disease      SNOMEDCT    101                      Disease   \n",
+       "55        PhenotypicFeature            HP    957                        Pheno   \n",
+       "56        PhenotypicFeature          NCIT    116                        Pheno   \n",
+       "57        PhenotypicFeature      SNOMEDCT      8                        Pheno   \n",
+       "58                Procedure          NCIT     13                    Procedure   \n",
+       "59                  Disease         MONDO   1007                      Disease   \n",
+       "60                  Disease          NCIT      5                      Disease   \n",
+       "61                  Disease      SNOMEDCT     81                      Disease   \n",
+       "62        PhenotypicFeature            HP    693                        Pheno   \n",
+       "63        PhenotypicFeature          NCIT     34                        Pheno   \n",
+       "64        PhenotypicFeature      SNOMEDCT      7                        Pheno   \n",
+       "65                Procedure          NCIT      2                    Procedure   \n",
+       "66                  Disease         MONDO  18759                      Disease   \n",
+       "67                  Disease          NCIT    113                      Disease   \n",
+       "68                  Disease      SNOMEDCT   1532                      Disease   \n",
+       "69        PhenotypicFeature            HP  11425                        Pheno   \n",
+       "70        PhenotypicFeature          NCIT   1257                        Pheno   \n",
+       "71        PhenotypicFeature      SNOMEDCT    121                        Pheno   \n",
+       "72                Procedure          NCIT    132                    Procedure   \n",
+       "73                  Disease         MONDO  12086                      Disease   \n",
+       "74                  Disease          NCIT     86                      Disease   \n",
+       "75                  Disease      SNOMEDCT   1055                      Disease   \n",
+       "76        PhenotypicFeature            HP  10355                        Pheno   \n",
+       "77        PhenotypicFeature          NCIT    733                        Pheno   \n",
+       "78        PhenotypicFeature      SNOMEDCT     78                        Pheno   \n",
+       "79                Procedure          NCIT     67                    Procedure   \n",
+       "80                  Disease         MONDO   1170                      Disease   \n",
+       "81                  Disease          NCIT      7                      Disease   \n",
+       "82                  Disease      SNOMEDCT     94                      Disease   \n",
+       "83        PhenotypicFeature            HP    705                        Pheno   \n",
+       "84        PhenotypicFeature          NCIT     56                        Pheno   \n",
+       "85        PhenotypicFeature      SNOMEDCT      9                        Pheno   \n",
+       "86                Procedure          NCIT      7                    Procedure   \n",
+       "87                  Disease         MONDO    380                      Disease   \n",
+       "88                  Disease          NCIT      3                      Disease   \n",
+       "89                  Disease      SNOMEDCT     36                      Disease   \n",
+       "90        PhenotypicFeature            HP    395                        Pheno   \n",
+       "91        PhenotypicFeature          NCIT     34                        Pheno   \n",
+       "92        PhenotypicFeature      SNOMEDCT      1                        Pheno   \n",
+       "93                Procedure          NCIT      3                    Procedure   \n",
+       "94                  Disease         MONDO    307                      Disease   \n",
+       "95                  Disease          NCIT      1                      Disease   \n",
+       "96                  Disease      SNOMEDCT     26                      Disease   \n",
+       "97        PhenotypicFeature            HP    217                        Pheno   \n",
+       "98        PhenotypicFeature          NCIT     25                        Pheno   \n",
+       "99        PhenotypicFeature      SNOMEDCT      2                        Pheno   \n",
+       "100               Procedure          NCIT      2                    Procedure   \n",
+       "101                 Disease         MONDO    158                      Disease   \n",
+       "102                 Disease          NCIT      2                      Disease   \n",
+       "103                 Disease      SNOMEDCT     13                      Disease   \n",
+       "104       PhenotypicFeature            HP    113                        Pheno   \n",
+       "105       PhenotypicFeature          NCIT      5                        Pheno   \n",
+       "106               Procedure          NCIT      1                    Procedure   \n",
+       "107                 Disease         MONDO    721                      Disease   \n",
+       "108                 Disease          NCIT      4                      Disease   \n",
+       "109                 Disease      SNOMEDCT     58                      Disease   \n",
+       "110       PhenotypicFeature            HP    502                        Pheno   \n",
+       "111       PhenotypicFeature          NCIT     56                        Pheno   \n",
+       "112       PhenotypicFeature      SNOMEDCT      4                        Pheno   \n",
+       "113               Procedure          NCIT      3                    Procedure   \n",
+       "114                 Disease         MONDO    209                      Disease   \n",
+       "115                 Disease          NCIT      2                      Disease   \n",
+       "116                 Disease      SNOMEDCT     20                      Disease   \n",
+       "117       PhenotypicFeature            HP    158                        Pheno   \n",
+       "118       PhenotypicFeature          NCIT      4                        Pheno   \n",
+       "119       PhenotypicFeature      SNOMEDCT      2                        Pheno   \n",
+       "120               Procedure          NCIT      2                    Procedure   \n",
+       "\n",
+       "    subject_category_postfix_abbv                       joined_to_match_xbte  \n",
+       "0                            Chem           ChemCHEBI_decreased_DiseaseMONDO  \n",
+       "1                            Chem            ChemCHEBI_decreased_DiseaseNCIT  \n",
+       "2                            Chem        ChemCHEBI_decreased_DiseaseSNOMEDCT  \n",
+       "3                            Chem                ChemCHEBI_decreased_PhenoHP  \n",
+       "4                            Chem              ChemCHEBI_decreased_PhenoNCIT  \n",
+       "5                            Chem          ChemCHEBI_decreased_PhenoSNOMEDCT  \n",
+       "6                            Chem          ChemCHEBI_decreased_ProcedureNCIT  \n",
+       "7                            Chem           ChemCHEBI_increased_DiseaseMONDO  \n",
+       "8                            Chem            ChemCHEBI_increased_DiseaseNCIT  \n",
+       "9                            Chem        ChemCHEBI_increased_DiseaseSNOMEDCT  \n",
+       "10                           Chem                ChemCHEBI_increased_PhenoHP  \n",
+       "11                           Chem              ChemCHEBI_increased_PhenoNCIT  \n",
+       "12                           Chem          ChemCHEBI_increased_PhenoSNOMEDCT  \n",
+       "13                           Chem          ChemCHEBI_increased_ProcedureNCIT  \n",
+       "14                           Chem            ChemUNII_decreased_DiseaseMONDO  \n",
+       "15                           Chem             ChemUNII_decreased_DiseaseNCIT  \n",
+       "16                           Chem         ChemUNII_decreased_DiseaseSNOMEDCT  \n",
+       "17                           Chem                 ChemUNII_decreased_PhenoHP  \n",
+       "18                           Chem               ChemUNII_decreased_PhenoNCIT  \n",
+       "19                           Chem           ChemUNII_decreased_PhenoSNOMEDCT  \n",
+       "20                           Chem           ChemUNII_decreased_ProcedureNCIT  \n",
+       "21                           Chem            ChemUNII_increased_DiseaseMONDO  \n",
+       "22                           Chem             ChemUNII_increased_DiseaseNCIT  \n",
+       "23                           Chem         ChemUNII_increased_DiseaseSNOMEDCT  \n",
+       "24                           Chem                 ChemUNII_increased_PhenoHP  \n",
+       "25                           Chem               ChemUNII_increased_PhenoNCIT  \n",
+       "26                           Chem           ChemUNII_increased_PhenoSNOMEDCT  \n",
+       "27                           Chem           ChemUNII_increased_ProcedureNCIT  \n",
+       "28                        Disease        DiseaseMONDO_decreased_DiseaseMONDO  \n",
+       "29                        Disease         DiseaseMONDO_decreased_DiseaseNCIT  \n",
+       "30                        Disease     DiseaseMONDO_decreased_DiseaseSNOMEDCT  \n",
+       "31                        Disease             DiseaseMONDO_decreased_PhenoHP  \n",
+       "32                        Disease           DiseaseMONDO_decreased_PhenoNCIT  \n",
+       "33                        Disease       DiseaseMONDO_decreased_PhenoSNOMEDCT  \n",
+       "34                        Disease       DiseaseMONDO_decreased_ProcedureNCIT  \n",
+       "35                        Disease        DiseaseMONDO_increased_DiseaseMONDO  \n",
+       "36                        Disease         DiseaseMONDO_increased_DiseaseNCIT  \n",
+       "37                        Disease     DiseaseMONDO_increased_DiseaseSNOMEDCT  \n",
+       "38                        Disease             DiseaseMONDO_increased_PhenoHP  \n",
+       "39                        Disease           DiseaseMONDO_increased_PhenoNCIT  \n",
+       "40                        Disease       DiseaseMONDO_increased_PhenoSNOMEDCT  \n",
+       "41                        Disease       DiseaseMONDO_increased_ProcedureNCIT  \n",
+       "42                        Disease         DiseaseNCIT_decreased_DiseaseMONDO  \n",
+       "43                        Disease      DiseaseNCIT_decreased_DiseaseSNOMEDCT  \n",
+       "44                        Disease              DiseaseNCIT_decreased_PhenoHP  \n",
+       "45                        Disease            DiseaseNCIT_decreased_PhenoNCIT  \n",
+       "46                        Disease        DiseaseNCIT_decreased_ProcedureNCIT  \n",
+       "47                        Disease         DiseaseNCIT_increased_DiseaseMONDO  \n",
+       "48                        Disease      DiseaseNCIT_increased_DiseaseSNOMEDCT  \n",
+       "49                        Disease              DiseaseNCIT_increased_PhenoHP  \n",
+       "50                        Disease            DiseaseNCIT_increased_PhenoNCIT  \n",
+       "51                        Disease        DiseaseNCIT_increased_PhenoSNOMEDCT  \n",
+       "52                        Disease     DiseaseSNOMEDCT_decreased_DiseaseMONDO  \n",
+       "53                        Disease      DiseaseSNOMEDCT_decreased_DiseaseNCIT  \n",
+       "54                        Disease  DiseaseSNOMEDCT_decreased_DiseaseSNOMEDCT  \n",
+       "55                        Disease          DiseaseSNOMEDCT_decreased_PhenoHP  \n",
+       "56                        Disease        DiseaseSNOMEDCT_decreased_PhenoNCIT  \n",
+       "57                        Disease    DiseaseSNOMEDCT_decreased_PhenoSNOMEDCT  \n",
+       "58                        Disease    DiseaseSNOMEDCT_decreased_ProcedureNCIT  \n",
+       "59                        Disease     DiseaseSNOMEDCT_increased_DiseaseMONDO  \n",
+       "60                        Disease      DiseaseSNOMEDCT_increased_DiseaseNCIT  \n",
+       "61                        Disease  DiseaseSNOMEDCT_increased_DiseaseSNOMEDCT  \n",
+       "62                        Disease          DiseaseSNOMEDCT_increased_PhenoHP  \n",
+       "63                        Disease        DiseaseSNOMEDCT_increased_PhenoNCIT  \n",
+       "64                        Disease    DiseaseSNOMEDCT_increased_PhenoSNOMEDCT  \n",
+       "65                        Disease    DiseaseSNOMEDCT_increased_ProcedureNCIT  \n",
+       "66                          Pheno             PhenoHP_decreased_DiseaseMONDO  \n",
+       "67                          Pheno              PhenoHP_decreased_DiseaseNCIT  \n",
+       "68                          Pheno          PhenoHP_decreased_DiseaseSNOMEDCT  \n",
+       "69                          Pheno                  PhenoHP_decreased_PhenoHP  \n",
+       "70                          Pheno                PhenoHP_decreased_PhenoNCIT  \n",
+       "71                          Pheno            PhenoHP_decreased_PhenoSNOMEDCT  \n",
+       "72                          Pheno            PhenoHP_decreased_ProcedureNCIT  \n",
+       "73                          Pheno             PhenoHP_increased_DiseaseMONDO  \n",
+       "74                          Pheno              PhenoHP_increased_DiseaseNCIT  \n",
+       "75                          Pheno          PhenoHP_increased_DiseaseSNOMEDCT  \n",
+       "76                          Pheno                  PhenoHP_increased_PhenoHP  \n",
+       "77                          Pheno                PhenoHP_increased_PhenoNCIT  \n",
+       "78                          Pheno            PhenoHP_increased_PhenoSNOMEDCT  \n",
+       "79                          Pheno            PhenoHP_increased_ProcedureNCIT  \n",
+       "80                          Pheno           PhenoNCIT_decreased_DiseaseMONDO  \n",
+       "81                          Pheno            PhenoNCIT_decreased_DiseaseNCIT  \n",
+       "82                          Pheno        PhenoNCIT_decreased_DiseaseSNOMEDCT  \n",
+       "83                          Pheno                PhenoNCIT_decreased_PhenoHP  \n",
+       "84                          Pheno              PhenoNCIT_decreased_PhenoNCIT  \n",
+       "85                          Pheno          PhenoNCIT_decreased_PhenoSNOMEDCT  \n",
+       "86                          Pheno          PhenoNCIT_decreased_ProcedureNCIT  \n",
+       "87                          Pheno           PhenoNCIT_increased_DiseaseMONDO  \n",
+       "88                          Pheno            PhenoNCIT_increased_DiseaseNCIT  \n",
+       "89                          Pheno        PhenoNCIT_increased_DiseaseSNOMEDCT  \n",
+       "90                          Pheno                PhenoNCIT_increased_PhenoHP  \n",
+       "91                          Pheno              PhenoNCIT_increased_PhenoNCIT  \n",
+       "92                          Pheno          PhenoNCIT_increased_PhenoSNOMEDCT  \n",
+       "93                          Pheno          PhenoNCIT_increased_ProcedureNCIT  \n",
+       "94                          Pheno       PhenoSNOMEDCT_decreased_DiseaseMONDO  \n",
+       "95                          Pheno        PhenoSNOMEDCT_decreased_DiseaseNCIT  \n",
+       "96                          Pheno    PhenoSNOMEDCT_decreased_DiseaseSNOMEDCT  \n",
+       "97                          Pheno            PhenoSNOMEDCT_decreased_PhenoHP  \n",
+       "98                          Pheno          PhenoSNOMEDCT_decreased_PhenoNCIT  \n",
+       "99                          Pheno      PhenoSNOMEDCT_decreased_PhenoSNOMEDCT  \n",
+       "100                         Pheno      PhenoSNOMEDCT_decreased_ProcedureNCIT  \n",
+       "101                         Pheno       PhenoSNOMEDCT_increased_DiseaseMONDO  \n",
+       "102                         Pheno        PhenoSNOMEDCT_increased_DiseaseNCIT  \n",
+       "103                         Pheno    PhenoSNOMEDCT_increased_DiseaseSNOMEDCT  \n",
+       "104                         Pheno            PhenoSNOMEDCT_increased_PhenoHP  \n",
+       "105                         Pheno          PhenoSNOMEDCT_increased_PhenoNCIT  \n",
+       "106                         Pheno      PhenoSNOMEDCT_increased_ProcedureNCIT  \n",
+       "107                     Procedure       ProcedureNCIT_decreased_DiseaseMONDO  \n",
+       "108                     Procedure        ProcedureNCIT_decreased_DiseaseNCIT  \n",
+       "109                     Procedure    ProcedureNCIT_decreased_DiseaseSNOMEDCT  \n",
+       "110                     Procedure            ProcedureNCIT_decreased_PhenoHP  \n",
+       "111                     Procedure          ProcedureNCIT_decreased_PhenoNCIT  \n",
+       "112                     Procedure      ProcedureNCIT_decreased_PhenoSNOMEDCT  \n",
+       "113                     Procedure      ProcedureNCIT_decreased_ProcedureNCIT  \n",
+       "114                     Procedure       ProcedureNCIT_increased_DiseaseMONDO  \n",
+       "115                     Procedure        ProcedureNCIT_increased_DiseaseNCIT  \n",
+       "116                     Procedure    ProcedureNCIT_increased_DiseaseSNOMEDCT  \n",
+       "117                     Procedure            ProcedureNCIT_increased_PhenoHP  \n",
+       "118                     Procedure          ProcedureNCIT_increased_PhenoNCIT  \n",
+       "119                     Procedure      ProcedureNCIT_increased_PhenoSNOMEDCT  \n",
+       "120                     Procedure      ProcedureNCIT_increased_ProcedureNCIT  "
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "# get a list of meta-triples for this KP: make it look like what BTE wants as closely as possible (see yaml)\n",
+    "# X_BTE KNOWN OPERATIONS TO ADD\n",
+    "#     unique combos of subject-prefix / subject-category / predicate / object-prefix / object-category \n",
+    "#     (and qualifier-set, if applicable)\n",
+    "\n",
+    "# Use the grouped[\"joined_to_match_xbte\"] column made below to create the tags and x-bte-kgs-operations portions in ehr_risk_kg.yaml\n",
+    "    \n",
+    "meta_triples = kg.copy()\n",
+    "\n",
+    "meta_triples[\"subject_prefix\"] = meta_triples[\"subject\"].str.split(\":\").str[0]\n",
+    "meta_triples[\"object_prefix\"] = meta_triples[\"object\"].str.split(\":\").str[0]\n",
+    "\n",
+    "meta_triples[\"subject_category_postfix\"] = meta_triples[\"subject_category\"].str.split(\":\").str[1]\n",
+    "meta_triples[\"object_category_postfix\"] = meta_triples[\"object_category\"].str.split(\":\").str[1]\n",
+    "\n",
+    "meta_triples['increased_or_decreased'] = None\n",
+    "meta_triples.loc[meta_triples['predicate'].str.contains('increased', case=False, na=False), 'increased_or_decreased'] = 'increased'\n",
+    "meta_triples.loc[meta_triples['predicate'].str.contains('decreased', case=False, na=False), 'increased_or_decreased'] = 'decreased'\n",
+    "\n",
+    "# unique combos of subject-prefix / subject-category / predicate / object-prefix / object-category\n",
+    "\n",
+    "# Define the columns you want to combine\n",
+    "columns_to_combine = ['subject_category_postfix', 'subject_prefix', 'increased_or_decreased', 'object_category_postfix', 'object_prefix']\n",
+    "\n",
+    "# Group by the combination of columns and count the occurrences\n",
+    "grouped = meta_triples.groupby(columns_to_combine).size().reset_index(name='Count')\n",
+    "grouped['object_category_postfix_abbv'] = grouped['object_category_postfix'].str.replace('PhenotypicFeature', 'Pheno')\n",
+    "grouped['subject_category_postfix_abbv'] = grouped['subject_category_postfix'].str.replace('PhenotypicFeature', 'Pheno')\n",
+    "\n",
+    "grouped['object_category_postfix_abbv'] = grouped['object_category_postfix_abbv'].str.replace('ChemicalEntity', 'Chem')\n",
+    "grouped['subject_category_postfix_abbv'] = grouped['subject_category_postfix_abbv'].str.replace('ChemicalEntity', 'Chem')\n",
+    "\n",
+    "grouped['joined_to_match_xbte'] = grouped['subject_category_postfix_abbv'].astype(str) + grouped['subject_prefix'].astype(str) + '_' + grouped['increased_or_decreased'].astype(str) + '_' + grouped['object_category_postfix_abbv'].astype(str) + grouped['object_prefix'].astype(str)\n",
+    "\n",
+    "# grouped.to_csv('yaml_all_metatriples.tsv', sep=\"\\t\")\n",
+    "# # Print the DataFrame with unique combinations and their counts\n",
+    "\n",
+    "with pd.option_context(\"display.max_rows\", 30000):\n",
+    "    display(grouped)\n",
+    "    "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 335,
+   "id": "ac11f711",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>subject</th>\n",
+       "      <th>predicate</th>\n",
+       "      <th>object</th>\n",
+       "      <th>relation</th>\n",
+       "      <th>provided_by</th>\n",
+       "      <th>provided_date</th>\n",
+       "      <th>predicate_category</th>\n",
+       "      <th>classifier</th>\n",
+       "      <th>auc_roc</th>\n",
+       "      <th>p_value</th>\n",
+       "      <th>...</th>\n",
+       "      <th>num_patients_without_condition</th>\n",
+       "      <th>nodes_frozenset</th>\n",
+       "      <th>subject_id</th>\n",
+       "      <th>subject_name</th>\n",
+       "      <th>subject_category</th>\n",
+       "      <th>object_id</th>\n",
+       "      <th>object_name</th>\n",
+       "      <th>object_category</th>\n",
+       "      <th>subject_prefix</th>\n",
+       "      <th>object_prefix</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>...</td>\n",
+       "      <td>9999902</td>\n",
+       "      <td>(HP:0000360, HP:0008629)</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>HP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>MONDO:0010643</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.999872</td>\n",
+       "      <td>...</td>\n",
+       "      <td>10000835</td>\n",
+       "      <td>(HP:0000360, MONDO:0010643)</td>\n",
+       "      <td>MONDO:0010643</td>\n",
+       "      <td>acute leukemia (disease)</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>HP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>UNII:25ADE2236L</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.936767</td>\n",
+       "      <td>...</td>\n",
+       "      <td>10000939</td>\n",
+       "      <td>(UNII:25ADE2236L, HP:0000360)</td>\n",
+       "      <td>UNII:25ADE2236L</td>\n",
+       "      <td>thrombin</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>UNII</td>\n",
+       "      <td>HP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>UNII:K16AIQ8CTM</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.998563</td>\n",
+       "      <td>...</td>\n",
+       "      <td>9998659</td>\n",
+       "      <td>(HP:0000360, UNII:K16AIQ8CTM)</td>\n",
+       "      <td>UNII:K16AIQ8CTM</td>\n",
+       "      <td>pertuzumab</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>UNII</td>\n",
+       "      <td>HP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>MONDO:0007972</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.840132</td>\n",
+       "      <td>0.093959</td>\n",
+       "      <td>...</td>\n",
+       "      <td>9998750</td>\n",
+       "      <td>(MONDO:0007972, HP:0000360)</td>\n",
+       "      <td>MONDO:0007972</td>\n",
+       "      <td>Meniere disease</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>HP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237099</th>\n",
+       "      <td>CHEBI:114785</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.998503</td>\n",
+       "      <td>...</td>\n",
+       "      <td>10007299</td>\n",
+       "      <td>(HP:0008629, CHEBI:114785)</td>\n",
+       "      <td>CHEBI:114785</td>\n",
+       "      <td>erlotinib</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>CHEBI</td>\n",
+       "      <td>HP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237100</th>\n",
+       "      <td>UNII:52CMI0WC3Y</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.999719</td>\n",
+       "      <td>...</td>\n",
+       "      <td>9996273</td>\n",
+       "      <td>(UNII:52CMI0WC3Y, HP:0008629)</td>\n",
+       "      <td>UNII:52CMI0WC3Y</td>\n",
+       "      <td>atezolizumab</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>UNII</td>\n",
+       "      <td>HP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237101</th>\n",
+       "      <td>CHEBI:135738</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.998357</td>\n",
+       "      <td>...</td>\n",
+       "      <td>9998030</td>\n",
+       "      <td>(HP:0008629, CHEBI:135738)</td>\n",
+       "      <td>CHEBI:135738</td>\n",
+       "      <td>clevidipine</td>\n",
+       "      <td>biolink:ChemicalEntity</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>CHEBI</td>\n",
+       "      <td>HP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237102</th>\n",
+       "      <td>MONDO:0004967</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.997631</td>\n",
+       "      <td>...</td>\n",
+       "      <td>10001385</td>\n",
+       "      <td>(HP:0008629, MONDO:0004967)</td>\n",
+       "      <td>MONDO:0004967</td>\n",
+       "      <td>acute lymphoblastic leukemia (disease)</td>\n",
+       "      <td>biolink:Disease</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>MONDO</td>\n",
+       "      <td>HP</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>237103</th>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>RO:0003308</td>\n",
+       "      <td>EHR Risk Provider (Multiomics)</td>\n",
+       "      <td>2022-05-18</td>\n",
+       "      <td>biolink:Association</td>\n",
+       "      <td>Logistic Regression</td>\n",
+       "      <td>0.959791</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>...</td>\n",
+       "      <td>9997731</td>\n",
+       "      <td>(HP:0000360, HP:0008629)</td>\n",
+       "      <td>HP:0000360</td>\n",
+       "      <td>Tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>HP:0008629</td>\n",
+       "      <td>Pulsatile tinnitus</td>\n",
+       "      <td>biolink:PhenotypicFeature</td>\n",
+       "      <td>HP</td>\n",
+       "      <td>HP</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>236812 rows × 23 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                subject                                        predicate  \\\n",
+       "0            HP:0008629  biolink:associated_with_increased_likelihood_of   \n",
+       "1         MONDO:0010643  biolink:associated_with_increased_likelihood_of   \n",
+       "2       UNII:25ADE2236L  biolink:associated_with_increased_likelihood_of   \n",
+       "3       UNII:K16AIQ8CTM  biolink:associated_with_increased_likelihood_of   \n",
+       "4         MONDO:0007972  biolink:associated_with_increased_likelihood_of   \n",
+       "...                 ...                                              ...   \n",
+       "237099     CHEBI:114785  biolink:associated_with_decreased_likelihood_of   \n",
+       "237100  UNII:52CMI0WC3Y  biolink:associated_with_decreased_likelihood_of   \n",
+       "237101     CHEBI:135738  biolink:associated_with_decreased_likelihood_of   \n",
+       "237102    MONDO:0004967  biolink:associated_with_decreased_likelihood_of   \n",
+       "237103       HP:0000360  biolink:associated_with_increased_likelihood_of   \n",
+       "\n",
+       "            object    relation                     provided_by provided_date  \\\n",
+       "0       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "1       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "2       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "3       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "4       HP:0000360  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "...            ...         ...                             ...           ...   \n",
+       "237099  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237100  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237101  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237102  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "237103  HP:0008629  RO:0003308  EHR Risk Provider (Multiomics)    2022-05-18   \n",
+       "\n",
+       "         predicate_category           classifier   auc_roc   p_value  ...  \\\n",
+       "0       biolink:Association  Logistic Regression  0.840132  0.000000  ...   \n",
+       "1       biolink:Association  Logistic Regression  0.840132  0.999872  ...   \n",
+       "2       biolink:Association  Logistic Regression  0.840132  0.936767  ...   \n",
+       "3       biolink:Association  Logistic Regression  0.840132  0.998563  ...   \n",
+       "4       biolink:Association  Logistic Regression  0.840132  0.093959  ...   \n",
+       "...                     ...                  ...       ...       ...  ...   \n",
+       "237099  biolink:Association  Logistic Regression  0.959791  0.998503  ...   \n",
+       "237100  biolink:Association  Logistic Regression  0.959791  0.999719  ...   \n",
+       "237101  biolink:Association  Logistic Regression  0.959791  0.998357  ...   \n",
+       "237102  biolink:Association  Logistic Regression  0.959791  0.997631  ...   \n",
+       "237103  biolink:Association  Logistic Regression  0.959791  0.000000  ...   \n",
+       "\n",
+       "        num_patients_without_condition                nodes_frozenset  \\\n",
+       "0                              9999902       (HP:0000360, HP:0008629)   \n",
+       "1                             10000835    (HP:0000360, MONDO:0010643)   \n",
+       "2                             10000939  (UNII:25ADE2236L, HP:0000360)   \n",
+       "3                              9998659  (HP:0000360, UNII:K16AIQ8CTM)   \n",
+       "4                              9998750    (MONDO:0007972, HP:0000360)   \n",
+       "...                                ...                            ...   \n",
+       "237099                        10007299     (HP:0008629, CHEBI:114785)   \n",
+       "237100                         9996273  (UNII:52CMI0WC3Y, HP:0008629)   \n",
+       "237101                         9998030     (HP:0008629, CHEBI:135738)   \n",
+       "237102                        10001385    (HP:0008629, MONDO:0004967)   \n",
+       "237103                         9997731       (HP:0000360, HP:0008629)   \n",
+       "\n",
+       "             subject_id                            subject_name  \\\n",
+       "0            HP:0008629                      Pulsatile tinnitus   \n",
+       "1         MONDO:0010643                acute leukemia (disease)   \n",
+       "2       UNII:25ADE2236L                                thrombin   \n",
+       "3       UNII:K16AIQ8CTM                              pertuzumab   \n",
+       "4         MONDO:0007972                         Meniere disease   \n",
+       "...                 ...                                     ...   \n",
+       "237099     CHEBI:114785                               erlotinib   \n",
+       "237100  UNII:52CMI0WC3Y                            atezolizumab   \n",
+       "237101     CHEBI:135738                             clevidipine   \n",
+       "237102    MONDO:0004967  acute lymphoblastic leukemia (disease)   \n",
+       "237103       HP:0000360                                Tinnitus   \n",
+       "\n",
+       "                 subject_category   object_id         object_name  \\\n",
+       "0       biolink:PhenotypicFeature  HP:0000360            Tinnitus   \n",
+       "1                 biolink:Disease  HP:0000360            Tinnitus   \n",
+       "2          biolink:ChemicalEntity  HP:0000360            Tinnitus   \n",
+       "3          biolink:ChemicalEntity  HP:0000360            Tinnitus   \n",
+       "4                 biolink:Disease  HP:0000360            Tinnitus   \n",
+       "...                           ...         ...                 ...   \n",
+       "237099     biolink:ChemicalEntity  HP:0008629  Pulsatile tinnitus   \n",
+       "237100     biolink:ChemicalEntity  HP:0008629  Pulsatile tinnitus   \n",
+       "237101     biolink:ChemicalEntity  HP:0008629  Pulsatile tinnitus   \n",
+       "237102            biolink:Disease  HP:0008629  Pulsatile tinnitus   \n",
+       "237103  biolink:PhenotypicFeature  HP:0008629  Pulsatile tinnitus   \n",
+       "\n",
+       "                  object_category subject_prefix object_prefix  \n",
+       "0       biolink:PhenotypicFeature             HP            HP  \n",
+       "1       biolink:PhenotypicFeature          MONDO            HP  \n",
+       "2       biolink:PhenotypicFeature           UNII            HP  \n",
+       "3       biolink:PhenotypicFeature           UNII            HP  \n",
+       "4       biolink:PhenotypicFeature          MONDO            HP  \n",
+       "...                           ...            ...           ...  \n",
+       "237099  biolink:PhenotypicFeature          CHEBI            HP  \n",
+       "237100  biolink:PhenotypicFeature           UNII            HP  \n",
+       "237101  biolink:PhenotypicFeature          CHEBI            HP  \n",
+       "237102  biolink:PhenotypicFeature          MONDO            HP  \n",
+       "237103  biolink:PhenotypicFeature             HP            HP  \n",
+       "\n",
+       "[236812 rows x 23 columns]"
+      ]
+     },
+     "execution_count": 335,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# need this (dataframe called kg_yaml) to lookup examples for testExamples (qInput and oneOutput) portion of yaml\n",
+    "# additionally need this generate portion of x-bte-response-mapping\n",
+    "\n",
+    "kg_yaml = kg.copy()\n",
+    "\n",
+    "kg_yaml[\"subject_prefix\"] = kg_yaml[\"subject\"].str.split(\":\").str[0]\n",
+    "kg_yaml[\"object_prefix\"] = kg_yaml[\"object\"].str.split(\":\").str[0]\n",
+    "\n",
+    "kg_yaml"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 339,
+   "id": "f5c776cc",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# the following code writes the tags and x-bte-kgs-operations of ehr_risk_kg.yaml to an output file titled 'x-bte-kgs-operations-response-mapping.txt'\n",
+    "# copy paste the output to the relevant section in the yaml\n",
+    "with open('x-bte-kgs-operations-response-mapping.txt', 'w+') as file:\n",
+    "    file.write(\"      tags:\" + \"\\n\")\n",
+    "    file.write(\"      - query:\" + \"\\n\")\n",
+    "    file.write(f\"      ## {len(grouped)} operations based on TSV of KG\" + \"\\n\")\n",
+    "    for index, row in grouped.iterrows():\n",
+    "        file.write(f\"      ## - {row['joined_to_match_xbte']}\" + \"\\n\")\n",
+    "    file.write(\"      x-bte-kgs-operations\" + \"\\n\")\n",
+    "    for index, row in grouped.iterrows(): \n",
+    "        file.write(f\"      - $ref: '#/components/x-bte-kgs-operations/{row['joined_to_match_xbte']}'\" + \"\\n\")\n",
+    "        file.write(f\"      - $ref: '#/components/x-bte-kgs-operations/{row['joined_to_match_xbte']}-rev'\" + \"\\n\")\n",
+    "        "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 340,
+   "id": "232ad5c1",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# the following code writes the x-bte-kgs-operations of ehr_risk_kg.yaml to an output file titled 'x-bte-kgs-operations-response-mapping.txt'\n",
+    "# copy paste the output to the relevant section in the yaml\n",
+    "\n",
+    "with open('x-bte-kgs-operations-response-mapping.txt', 'a') as file:\n",
+    "    file.write(\"  \" + \"x-bte-kgs-operations:\" + \"\\n\")\n",
+    "    for index, row in grouped.iterrows():\n",
+    "        xbte_header = f'{row[\"joined_to_match_xbte\"]}:'\n",
+    "        file.write(\"    \" + xbte_header + \"\\n\")\n",
+    "        \n",
+    "        base_url = \"https://pending.biothings.io/multiomics_ehr_risk_kp/query?q=\"\n",
+    "        url = base_url + f'''subject.type:\"biolink:{row['subject_category_postfix']}\"%20AND%20_exists_:subject.\"biolink:{row['subject_prefix']}\"%20AND%20association.predicate:associated_with_decreased_likelihood_of%20AND%20object.type:\"biolink:{row['object_category_postfix']}\"%20AND%20_exists_:object.{row['object_prefix']}'''\n",
+    "        commented_url = \"## \" + url\n",
+    "        file.write(\"    \" + commented_url + \"\\n\")\n",
+    "        \n",
+    "        file.write(\"    ## \" + str(row[\"Count\"]) + \" records\" + \"\\n\")\n",
+    "        \n",
+    "        file.write(\"      -supportBatch: true\" + \"\\n\")\n",
+    "        file.write(\"       useTemplating: true\" + \"\\n\")\n",
+    "        file.write(\"       inputs:\" + \"\\n\")\n",
+    "        \n",
+    "        file.write(f\"          - id: {row['subject_prefix']}\" + \"\\n\")\n",
+    "        file.write(f\"          - semantic: {row['subject_category_postfix']}\" + \"\\n\")\n",
+    "        \n",
+    "        file.write(\"       requestBody:\" + \"\\n\")\n",
+    "        file.write(\"         body: >-\" + \"\\n\")\n",
+    "        \n",
+    "        file.write(\"           {\" + \"\\n\")\n",
+    "        q_string = f\"\"\" \"q\": [ {{ queryInputs | rmPrefix() | wrap ( '[\"' , '\",\"biolink:{row['subject_category_postfix']}\", \"associated_with_{row['increased_or_decreased']}_likelihood_of\", \"biolink:{row['object_category_postfix']}\"]') }} ]\"\"\"\n",
+    "        file.write('              ' + q_string + \"\\n\") \n",
+    "        scopes_string = f\"\"\" \"scopes\": [\"subject.{row['subject_prefix']}\", \"subject.type\", \"association.predicate\", \"object.type\"]\"\"\"\n",
+    "        file.write('              ' + scopes_string + \"\\n\") \n",
+    "        file.write(\"           }\" + \"\\n\")\n",
+    "        \n",
+    "        file.write(\"       parameters:\" + \"\\n\")\n",
+    "        file.write(f\"         fields: object.{row['object_prefix']}, association.edge_attributes, subject.name, object.name, source.edge_sources\" + \"\\n\")\n",
+    "        file.write(f\"         size: 1000\" + \"\\n\")\n",
+    "        \n",
+    "        file.write(\"       outputs:\" + \"\\n\")\n",
+    "        file.write(f\"         -id: {row['object_prefix']}\" + \"\\n\")\n",
+    "        file.write(f\"          semantic: {row['object_category_postfix']}\" + \"\\n\")\n",
+    "        \n",
+    "        file.write(\"       predicate: associated_with\" + \"\\n\")\n",
+    "        file.write(\"       qualifiers:\" + \"\\n\")\n",
+    "        file.write(f\"         object_direction_qualifier: {row['increased_or_decreased']}\" + \"\\n\")\n",
+    "        file.write(f\"         object_aspect_qualifier: likelihood\" + \"\\n\")\n",
+    "        \n",
+    "        file.write(\"       response_mapping:\" + \"\\n\")\n",
+    "        file.write(f'         \"ref\": \"#/components/x-bte-response-mapping/object-{row[\"object_prefix\"]}\"' + \"\\n\")\n",
+    "        \n",
+    "        file.write(\"       # testExamples:\" + \"\\n\")\n",
+    "        sub_prefix = row[\"subject_prefix\"]\n",
+    "        obj_prefix = row[\"object_prefix\"]\n",
+    "        matched_rows = kg_yaml.loc[(kg_yaml['subject_prefix'] == sub_prefix) & (kg_yaml['object_prefix'] == obj_prefix)]\n",
+    "        select_row = matched_rows.sample()\n",
+    "        qInput = select_row['subject'].values[0]\n",
+    "        qName = select_row['subject_name'].values[0]\n",
+    "        oneOutput = select_row['object'].values[0]\n",
+    "        oneOutputName = select_row['object_name'].values[0]\n",
+    "        file.write(f'       #   - qInput: \"{qInput}\"      ## {qName} ' + \"\\n\")\n",
+    "        file.write(f'       #     oneOutput: \"{oneOutput}\"      ## {oneOutputName} ' + \"\\n\")\n",
+    "        \n",
+    "        #### DO REVERSE #### \n",
+    "        \n",
+    "        xbte_header_rev = f'{row[\"joined_to_match_xbte\"]}_rev:'\n",
+    "        file.write(\"    \" + xbte_header_rev + \"\\n\")\n",
+    "\n",
+    "        file.write(\"      -supportBatch: true\" + \"\\n\")\n",
+    "        file.write(\"       useTemplating: true\" + \"\\n\")\n",
+    "        file.write(\"       inputs:\" + \"\\n\")\n",
+    "        \n",
+    "        file.write(f\"          - id: {row['object_prefix']}\" + \"\\n\")\n",
+    "        file.write(f\"          - semantic: {row['object_category_postfix']}\" + \"\\n\")\n",
+    "        \n",
+    "        file.write(\"       requestBody:\" + \"\\n\")\n",
+    "        file.write(\"         body: >-\" + \"\\n\")\n",
+    "        \n",
+    "        file.write(\"           {\" + \"\\n\")\n",
+    "        q_string = f\"\"\" \"q\": [ {{ queryInputs | rmPrefix() | wrap ( '[\"' , '\",\"biolink:{row['subject_category_postfix']}\", \"associated_with_{row['increased_or_decreased']}_likelihood_of\", \"biolink:{row['object_category_postfix']}\"]') }} ]\"\"\"\n",
+    "        file.write('              ' + q_string + \"\\n\") \n",
+    "        scopes_string = f\"\"\" \"scopes\": [\"object.{row['object_prefix']}\", \"subject.type\", \"association.predicate\", \"object.type\"]\"\"\"\n",
+    "        file.write('              ' + scopes_string + \"\\n\") \n",
+    "        file.write(\"           }\" + \"\\n\")\n",
+    "        \n",
+    "        file.write(\"       parameters:\" + \"\\n\")\n",
+    "        file.write(f\"         fields: subject.{row['subject_prefix']}, association.edge_attributes, subject.name, object.name, source.edge_sources\" + \"\\n\")\n",
+    "        file.write(f\"         size: 1000\" + \"\\n\")\n",
+    "        \n",
+    "        file.write(\"       outputs:\" + \"\\n\")\n",
+    "        file.write(f\"         -id: {row['subject_prefix']}\" + \"\\n\")\n",
+    "        file.write(f\"          semantic: {row['subject_category_postfix']}\" + \"\\n\")\n",
+    "        \n",
+    "        file.write(\"       predicate: associated_with\" + \"\\n\")\n",
+    "        file.write(\"       qualifiers:\" + \"\\n\")\n",
+    "        file.write(f\"         object_direction_qualifier: {row['increased_or_decreased']}\" + \"\\n\")\n",
+    "        file.write(f\"         object_aspect_qualifier: likelihood\" + \"\\n\")\n",
+    "        \n",
+    "        file.write(\"       response_mapping:\" + \"\\n\")\n",
+    "        file.write(f'         \"ref\": \"#/components/x-bte-response-mapping/object-{row[\"subject_prefix\"]}\"' + \"\\n\")\n",
+    "        \n",
+    "        file.write(\"       # testExamples:\" + \"\\n\")\n",
+    "        sub_prefix = row[\"subject_prefix\"]\n",
+    "        obj_prefix = row[\"object_prefix\"]\n",
+    "        matched_rows = kg_yaml.loc[(kg_yaml['subject_prefix'] == sub_prefix) & (kg_yaml['object_prefix'] == obj_prefix)]\n",
+    "        select_row = matched_rows.sample()\n",
+    "        qInput = select_row['object'].values[0]\n",
+    "        qName = select_row['object_name'].values[0]\n",
+    "        oneOutput = select_row['subject'].values[0]\n",
+    "        oneOutputName = select_row['subject_name'].values[0]\n",
+    "        file.write(f'       #   - qInput: \"{qInput}\"      ## {qName} ' + \"\\n\")\n",
+    "        file.write(f'       #     oneOutput: \"{oneOutput}\"      ## {oneOutputName} ' + \"\\n\")\n",
+    "        \n",
+    "        \n",
+    "\n",
+    "    "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "9bbc605f",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "Output example for x-bte-kgs-operations\n",
+    "\n",
+    "    ChemCHEBI_decreased_PhenoNCIT:\n",
+    "    ## https://pending.biothings.io/multiomics_ehr_risk_kp/query?q=subject.type:ChemicalEntity%20AND%20_exists_:subject.CHEBI%20AND%20association.predicate:associated_with_decreased_likelihood_of%20AND%20object.type:PhenotypicFeature%20AND%20_exists_:object.NCIT\n",
+    "    ## 191 records\n",
+    "      - supportBatch: true\n",
+    "        useTemplating: true  \n",
+    "        inputs:\n",
+    "          - id: CHEBI\n",
+    "            semantic: SmallMolecule\n",
+    "        requestBodyType: object\n",
+    "        requestBody:\n",
+    "          body: >-\n",
+    "            {\n",
+    "              \"q\": [ {{ queryInputs | rmPrefix() | wrap ( '[\"' , '\",\"biolink:ChemicalEntity\", \"associated_with_decreased_likelihood_of\", \"biolink:PhenotypicFeature\"]') }} ],\n",
+    "              \"scopes\": [\"subject.CHEBI\", \"subject.type\", \"association.predicate\", \"object.type\"]\n",
+    "            }\n",
+    "        parameters:\n",
+    "          fields: >-\n",
+    "            object.NCIT,\n",
+    "            association.edge_attributes,\n",
+    "            subject.name,object.name\n",
+    "          size: 1000\n",
+    "        outputs:\n",
+    "          - id: NCIT\n",
+    "            semantic: PhenotypicFeature\n",
+    "        predicate: associated_with\n",
+    "        qualifiers:\n",
+    "          object_direction_qualifier: decreased\n",
+    "          object_aspect_qualifier: likelihood\n",
+    "        response_mapping:\n",
+    "          \"$ref\": \"#/components/x-bte-response-mapping/object-NCIT\"\n",
+    "        # testExamples:\n",
+    "        #   - qInput: \"CHEBI:135866\"        ## clindamycin\n",
+    "        #     oneOutput: \"NCIT:C171647\"     ## SARS Coronavirus 2 Positive\n",
+    "    ChemCHEBI_decreased_PhenoNCIT-rev:\n",
+    "      - supportBatch: true\n",
+    "        useTemplating: true  \n",
+    "        inputs:\n",
+    "          - id: NCIT\n",
+    "            semantic: PhenotypicFeature\n",
+    "        requestBodyType: object\n",
+    "        requestBody:\n",
+    "          body: >-\n",
+    "            {\n",
+    "              \"q\": [ {{ queryInputs | rmPrefix() | wrap( '[\"' , '\",\"biolink:ChemicalEntity\", \"associated_with_decreased_likelihood_of\", \"biolink:PhenotypicFeature\"]') }} ],\n",
+    "              \"scopes\": [\"object.NCIT\", \"subject.type\", \"association.predicate\", \"object.type\"]\n",
+    "            }\n",
+    "        parameters:\n",
+    "          fields: >-\n",
+    "            subject.CHEBI,\n",
+    "            association.edge_attributes,\n",
+    "            subject.name,object.name\n",
+    "          size: 1000\n",
+    "        outputs:\n",
+    "          - id: CHEBI\n",
+    "            semantic: SmallMolecule\n",
+    "        predicate: associated_with\n",
+    "        qualifiers:\n",
+    "          subject_direction_qualifier: decreased\n",
+    "          subject_aspect_qualifier: likelihood\n",
+    "        response_mapping:\n",
+    "          \"$ref\": \"#/components/x-bte-response-mapping/subject-CHEBI\"\n",
+    "        # testExamples:\n",
+    "        #   - qInput: \"NCIT:C171647\"          ## SARS Coronavirus 2 Positive\n",
+    "        #     oneOutput: \"CHEBI:119915\"       ## fentanyl"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 341,
+   "id": "d834fe3e",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# the following code writes the x-bte-response-maping of ehr_risk_kg.yaml to an output file titled 'x-bte-kgs-operations-response-mapping.txt'\n",
+    "# copy paste the output to the relevant section in the yaml\n",
+    "with open('x-bte-kgs-operations-response-mapping.txt', 'a') as file:\n",
+    "    file.write(\"  \" + \"x-bte-response-mapping:\" + \"\\n\")\n",
+    "    for sprefix in list(kg_yaml['subject_prefix'].unique()):\n",
+    "        file.write(f\"    subject-{sprefix}:\" + \"\\n\")\n",
+    "        file.write(f\"      {sprefix}: subject.{sprefix}\" + \"\\n\")\n",
+    "        file.write(f\"      input_name: subject.name\" + \"\\n\")\n",
+    "        file.write(f\"      output_name: object.name\" + \"\\n\")\n",
+    "        file.write(f\"      edge-attributes: sources.edge_attributes\" + \"\\n\")\n",
+    "        file.write(f\"      trapi_sources: source.edge_sources\" + \"\\n\")\n",
+    "    for oprefix in list(kg_yaml['object_prefix'].unique()):\n",
+    "        file.write(f\"    object-{oprefix}:\" + \"\\n\")\n",
+    "        file.write(f\"      {oprefix}: object.{oprefix}\" + \"\\n\")\n",
+    "        file.write(f\"      input_name: object.name\" + \"\\n\")\n",
+    "        file.write(f\"      output_name: subject.name\" + \"\\n\")\n",
+    "        file.write(f\"      edge-attributes: sources.edge_attributes\" + \"\\n\")\n",
+    "        file.write(f\"      trapi_sources: source.edge_sources\" + \"\\n\")\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 325,
+   "id": "30781582",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>subject</th>\n",
+       "      <th>object</th>\n",
+       "      <th>predicate</th>\n",
+       "      <th>count</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>111072</th>\n",
+       "      <td>HP:0001974</td>\n",
+       "      <td>NCIT:C167118</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>4</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>211014</th>\n",
+       "      <td>NCIT:C167118</td>\n",
+       "      <td>HP:0001974</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>4</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>117058</th>\n",
+       "      <td>HP:0002900</td>\n",
+       "      <td>NCIT:C167118</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>4</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>122579</th>\n",
+       "      <td>HP:0003124</td>\n",
+       "      <td>NCIT:C167118</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>4</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>117810</th>\n",
+       "      <td>HP:0002902</td>\n",
+       "      <td>NCIT:C167118</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>3</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>78133</th>\n",
+       "      <td>CHEBI:77431</td>\n",
+       "      <td>MONDO:0004721</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>1</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>78134</th>\n",
+       "      <td>CHEBI:77431</td>\n",
+       "      <td>MONDO:0004866</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>1</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>78135</th>\n",
+       "      <td>CHEBI:77431</td>\n",
+       "      <td>MONDO:0004868</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>1</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>78136</th>\n",
+       "      <td>CHEBI:77431</td>\n",
+       "      <td>MONDO:0004946</td>\n",
+       "      <td>biolink:associated_with_increased_likelihood_of</td>\n",
+       "      <td>1</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>234248</th>\n",
+       "      <td>UNII:X85G7936GV</td>\n",
+       "      <td>SNOMEDCT:76571007</td>\n",
+       "      <td>biolink:associated_with_decreased_likelihood_of</td>\n",
+       "      <td>1</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>234249 rows × 4 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                subject             object  \\\n",
+       "111072       HP:0001974       NCIT:C167118   \n",
+       "211014     NCIT:C167118         HP:0001974   \n",
+       "117058       HP:0002900       NCIT:C167118   \n",
+       "122579       HP:0003124       NCIT:C167118   \n",
+       "117810       HP:0002902       NCIT:C167118   \n",
+       "...                 ...                ...   \n",
+       "78133       CHEBI:77431      MONDO:0004721   \n",
+       "78134       CHEBI:77431      MONDO:0004866   \n",
+       "78135       CHEBI:77431      MONDO:0004868   \n",
+       "78136       CHEBI:77431      MONDO:0004946   \n",
+       "234248  UNII:X85G7936GV  SNOMEDCT:76571007   \n",
+       "\n",
+       "                                              predicate  count  \n",
+       "111072  biolink:associated_with_decreased_likelihood_of      4  \n",
+       "211014  biolink:associated_with_decreased_likelihood_of      4  \n",
+       "117058  biolink:associated_with_increased_likelihood_of      4  \n",
+       "122579  biolink:associated_with_decreased_likelihood_of      4  \n",
+       "117810  biolink:associated_with_increased_likelihood_of      3  \n",
+       "...                                                 ...    ...  \n",
+       "78133   biolink:associated_with_decreased_likelihood_of      1  \n",
+       "78134   biolink:associated_with_decreased_likelihood_of      1  \n",
+       "78135   biolink:associated_with_increased_likelihood_of      1  \n",
+       "78136   biolink:associated_with_increased_likelihood_of      1  \n",
+       "234248  biolink:associated_with_decreased_likelihood_of      1  \n",
+       "\n",
+       "[234249 rows x 4 columns]"
+      ]
+     },
+     "execution_count": 325,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# # check to see how many identical triples there are in KG (requested by BTE)\n",
+    "# identical_triples_check = kg.groupby(['subject','object', 'predicate']).size().reset_index().rename(columns={0:'count'}).sort_values(by=['count'], ascending=False)\n",
+    "# identical_triples_check"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "647beb4b",
+   "metadata": {},
+   "source": [
+    "# Make a cleaned version of the parser below"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 32,
+   "id": "9a68e6a8",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{\n",
+      "  \"_id\": \"HP:0008629_HP:0000360_0.8401321539277617_00_8796399245685702_10102731\",\n",
+      "  \"subject\": {\n",
+      "    \"HP\": \"0008629\",\n",
+      "    \"id\": \"HP:0008629\",\n",
+      "    \"name\": \"Pulsatile tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.0,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 8.796399245685702,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10102731,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"MONDO:0010643_HP:0000360_0.8401321539277617_09998721067797812_8585212287149526_10107468\",\n",
+      "  \"subject\": {\n",
+      "    \"MONDO\": \"0010643\",\n",
+      "    \"id\": \"MONDO:0010643\",\n",
+      "    \"name\": \"acute leukemia (disease)\",\n",
+      "    \"type\": \"biolink:Disease\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.9998721067797812,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 8.585212287149526,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10107468,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"UNII:25ADE2236L_HP:0000360_0.8401321539277617_09367666401584368_4558176672832635_10095297\",\n",
+      "  \"subject\": {\n",
+      "    \"UNII\": \"25ADE2236L\",\n",
+      "    \"id\": \"UNII:25ADE2236L\",\n",
+      "    \"name\": \"thrombin\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.9367666401584368,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 4.558176672832635,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10095297,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"UNII:K16AIQ8CTM_HP:0000360_0.8401321539277617_09985626800193924_43575215395209606_10099409\",\n",
+      "  \"subject\": {\n",
+      "    \"UNII\": \"K16AIQ8CTM\",\n",
+      "    \"id\": \"UNII:K16AIQ8CTM\",\n",
+      "    \"name\": \"pertuzumab\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.9985626800193924,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 4.3575215395209606,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10099409,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"MONDO:0007972_HP:0000360_0.8401321539277617_009395878968875304_392606416950393_10100235\",\n",
+      "  \"subject\": {\n",
+      "    \"MONDO\": \"0007972\",\n",
+      "    \"id\": \"MONDO:0007972\",\n",
+      "    \"name\": \"Meniere disease\",\n",
+      "    \"type\": \"biolink:Disease\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.09395878968875304,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 3.92606416950393,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10100235,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"MONDO:0004866_HP:0000360_0.8401321539277617_00_30228399397470613_10094256\",\n",
+      "  \"subject\": {\n",
+      "    \"MONDO\": \"0004866\",\n",
+      "    \"id\": \"MONDO:0004866\",\n",
+      "    \"name\": \"eustachian tube disease\",\n",
+      "    \"type\": \"biolink:Disease\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.0,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 3.0228399397470613,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10094256,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"HP:0002321_HP:0000360_0.8401321539277617_00_297792187563902_10105748\",\n",
+      "  \"subject\": {\n",
+      "    \"HP\": \"0002321\",\n",
+      "    \"id\": \"HP:0002321\",\n",
+      "    \"name\": \"Vertigo\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.0,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 2.97792187563902,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10105748,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"HP:0011897_HP:0000360_0.8401321539277617_09882157516627652_29664217619814317_10103231\",\n",
+      "  \"subject\": {\n",
+      "    \"HP\": \"0011897\",\n",
+      "    \"id\": \"HP:0011897\",\n",
+      "    \"name\": \"Neutrophilia\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.9882157516627652,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 2.9664217619814317,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10103231,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"CHEBI:3403_HP:0000360_0.8401321539277617_07343407676622777_26320323984271305_10094497\",\n",
+      "  \"subject\": {\n",
+      "    \"CHEBI\": \"3403\",\n",
+      "    \"id\": \"CHEBI:3403\",\n",
+      "    \"name\": \"carboprost\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.7343407676622777,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 2.6320323984271305,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10094497,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"CHEBI:42758_HP:0000360_0.8401321539277617_041935921163506107_25903432895291627_10090857\",\n",
+      "  \"subject\": {\n",
+      "    \"CHEBI\": \"42758\",\n",
+      "    \"id\": \"CHEBI:42758\",\n",
+      "    \"name\": \"dextrose\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.41935921163506107,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 2.5903432895291627,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10090857,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"HP:0030788_HP:0000360_0.8401321539277617_00_2533460251458441_10098414\",\n",
+      "  \"subject\": {\n",
+      "    \"HP\": \"0030788\",\n",
+      "    \"id\": \"HP:0030788\",\n",
+      "    \"name\": \"Impacted cerumen\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.0,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 2.533460251458441,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10098414,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"MONDO:0012883_HP:0000360_0.8401321539277617_099451727004904_23595369689943997_10098573\",\n",
+      "  \"subject\": {\n",
+      "    \"MONDO\": \"0012883\",\n",
+      "    \"id\": \"MONDO:0012883\",\n",
+      "    \"name\": \"acute promyelocytic leukemia\",\n",
+      "    \"type\": \"biolink:Disease\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.99451727004904,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 2.3595369689943997,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10098573,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"CHEBI:15882_HP:0000360_0.8401321539277617_09962946928940472_21934060641992064_10099906\",\n",
+      "  \"subject\": {\n",
+      "    \"CHEBI\": \"15882\",\n",
+      "    \"id\": \"CHEBI:15882\",\n",
+      "    \"name\": \"phenol\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.9962946928940472,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 2.1934060641992064,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10099906,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"MONDO:0019065_HP:0000360_0.8401321539277617_09965366860113084_1909472317873171_10094912\",\n",
+      "  \"subject\": {\n",
+      "    \"MONDO\": \"0019065\",\n",
+      "    \"id\": \"MONDO:0019065\",\n",
+      "    \"name\": \"amyloidosis (disease)\",\n",
+      "    \"type\": \"biolink:Disease\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.9965366860113084,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.909472317873171,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10094912,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"UNII:1RXS4UE564_HP:0000360_0.8401321539277617_09145182816982448_17682389618102223_10099640\",\n",
+      "  \"subject\": {\n",
+      "    \"UNII\": \"1RXS4UE564\",\n",
+      "    \"id\": \"UNII:1RXS4UE564\",\n",
+      "    \"name\": \"alteplase\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.9145182816982448,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.7682389618102223,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10099640,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"CHEBI:2637_HP:0000360_0.8401321539277617_09984753422140926_17273309005514552_10097751\",\n",
+      "  \"subject\": {\n",
+      "    \"CHEBI\": \"2637\",\n",
+      "    \"id\": \"CHEBI:2637\",\n",
+      "    \"name\": \"amikacin\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.9984753422140926,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.7273309005514552,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10097751,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"CHEBI:28001_HP:0000360_0.8401321539277617_0327274960009581_17001729126903409_10099157\",\n",
+      "  \"subject\": {\n",
+      "    \"CHEBI\": \"28001\",\n",
+      "    \"id\": \"CHEBI:28001\",\n",
+      "    \"name\": \"vancomycin\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.327274960009581,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.7001729126903409,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10099157,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"CHEBI:32142_HP:0000360_0.8401321539277617_08551368959397363_16756681953392343_10100133\",\n",
+      "  \"subject\": {\n",
+      "    \"CHEBI\": \"32142\",\n",
+      "    \"id\": \"CHEBI:32142\",\n",
+      "    \"name\": \"sodium citrate\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.8551368959397363,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.6756681953392343,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10100133,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"CHEBI:5280_HP:0000360_0.8401321539277617_09843217708630094_16666971285058136_10101665\",\n",
+      "  \"subject\": {\n",
+      "    \"CHEBI\": \"5280\",\n",
+      "    \"id\": \"CHEBI:5280\",\n",
+      "    \"name\": \"gatifloxacin\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.9843217708630094,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.6666971285058136,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10101665,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"CHEBI:31705_HP:0000360_0.8401321539277617_09099121721955136_16359973452904688_10097430\",\n",
+      "  \"subject\": {\n",
+      "    \"CHEBI\": \"31705\",\n",
+      "    \"id\": \"CHEBI:31705\",\n",
+      "    \"name\": \"iodixanol\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.9099121721955136,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.6359973452904688,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10097430,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"HP:0002315_HP:0000360_0.8401321539277617_00_16166497779176117_10105844\",\n",
+      "  \"subject\": {\n",
+      "    \"HP\": \"0002315\",\n",
+      "    \"id\": \"HP:0002315\",\n",
+      "    \"name\": \"Headache\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.0,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.6166497779176117,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10105844,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"CHEBI:6709_HP:0000360_0.8401321539277617_9059419880941276e-14_1598480857062902_10100577\",\n",
+      "  \"subject\": {\n",
+      "    \"CHEBI\": \"6709\",\n",
+      "    \"id\": \"CHEBI:6709\",\n",
+      "    \"name\": \"meclizine\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 9.059419880941276e-14,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.598480857062902,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10100577,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"UNII:1BJ477IO2L_HP:0000360_0.8401321539277617_00_15423681686921502_10101289\",\n",
+      "  \"subject\": {\n",
+      "    \"UNII\": \"1BJ477IO2L\",\n",
+      "    \"id\": \"UNII:1BJ477IO2L\",\n",
+      "    \"name\": \"gadobutrol\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.0,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.5423681686921502,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10101289,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"CHEBI:9421_HP:0000360_0.8401321539277617_08516479806885544_14807202351994573_10096948\",\n",
+      "  \"subject\": {\n",
+      "    \"CHEBI\": \"9421\",\n",
+      "    \"id\": \"CHEBI:9421\",\n",
+      "    \"name\": \"tazobactam\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.8516479806885544,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.4807202351994573,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10096948,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"CHEBI:478164_HP:0000360_0.8401321539277617_07859646244357879_1454940814532088_10106536\",\n",
+      "  \"subject\": {\n",
+      "    \"CHEBI\": \"478164\",\n",
+      "    \"id\": \"CHEBI:478164\",\n",
+      "    \"name\": \"cefepime\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.7859646244357879,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.454940814532088,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10106536,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"CHEBI:4462_HP:0000360_0.8401321539277617_08450305750308531_13343018079195257_10093475\",\n",
+      "  \"subject\": {\n",
+      "    \"CHEBI\": \"4462\",\n",
+      "    \"id\": \"CHEBI:4462\",\n",
+      "    \"name\": \"sodium phosphate\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.8450305750308531,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.3343018079195257,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10093475,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"SNOMEDCT:127295002_HP:0000360_0.8401321539277617_05945757673388871_13294623195442723_10100698\",\n",
+      "  \"subject\": {\n",
+      "    \"SNOMEDCT\": \"127295002\",\n",
+      "    \"id\": \"SNOMEDCT:127295002\",\n",
+      "    \"name\": \"Traumatic brain injury (disorder)\",\n",
+      "    \"type\": \"biolink:Disease\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.5945757673388871,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.3294623195442723,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10100698,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"MONDO:0010030_HP:0000360_0.8401321539277617_08231940924433323_13284310019939671_10099754\",\n",
+      "  \"subject\": {\n",
+      "    \"MONDO\": \"0010030\",\n",
+      "    \"id\": \"MONDO:0010030\",\n",
+      "    \"name\": \"Sjogren syndrome\",\n",
+      "    \"type\": \"biolink:Disease\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.8231940924433323,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.3284310019939671,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10099754,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"UNII:FQ3DRG0N5K_HP:0000360_0.8401321539277617_09997741237977732_12685220499445722_10100132\",\n",
+      "  \"subject\": {\n",
+      "    \"UNII\": \"FQ3DRG0N5K\",\n",
+      "    \"id\": \"UNII:FQ3DRG0N5K\",\n",
+      "    \"name\": \"pancrelipase\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.9997741237977732,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.2685220499445722,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10100132,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"HP:0032372_HP:0000360_0.8401321539277617_09739456841856716_12651078229598214_10095958\",\n",
+      "  \"subject\": {\n",
+      "    \"HP\": \"0032372\",\n",
+      "    \"id\": \"HP:0032372\",\n",
+      "    \"name\": \"Increased peripheral blast count\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.9739456841856716,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.2651078229598214,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10095958,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"CHEBI:6754_HP:0000360_0.8401321539277617_05082567984079447_11344123128490888_10097573\",\n",
+      "  \"subject\": {\n",
+      "    \"CHEBI\": \"6754\",\n",
+      "    \"id\": \"CHEBI:6754\",\n",
+      "    \"name\": \"meperidine\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.5082567984079447,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.1344123128490888,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10097573,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"MONDO:0005391_HP:0000360_0.8401321539277617_019858432436236395_11125118199146522_10096019\",\n",
+      "  \"subject\": {\n",
+      "    \"MONDO\": \"0005391\",\n",
+      "    \"id\": \"MONDO:0005391\",\n",
+      "    \"name\": \"restless legs syndrome\",\n",
+      "    \"type\": \"biolink:Disease\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.19858432436236395,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.1125118199146522,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10096019,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"MONDO:0018935_HP:0000360_0.8401321539277617_099536202601544_11117474952487418_10099681\",\n",
+      "  \"subject\": {\n",
+      "    \"MONDO\": \"0018935\",\n",
+      "    \"id\": \"MONDO:0018935\",\n",
+      "    \"name\": \"hairy cell leukemia\",\n",
+      "    \"type\": \"biolink:Disease\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.99536202601544,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.1117474952487418,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10099681,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"MONDO:0011786_HP:0000360_0.8401321539277617_00_11102871262689296_10101809\",\n",
+      "  \"subject\": {\n",
+      "    \"MONDO\": \"0011786\",\n",
+      "    \"id\": \"MONDO:0011786\",\n",
+      "    \"name\": \"allergic rhinitis\",\n",
+      "    \"type\": \"biolink:Disease\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.0,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.1102871262689296,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10101809,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"MONDO:0002039_HP:0000360_0.8401321539277617_08357342064986439_10945588244442226_10102298\",\n",
+      "  \"subject\": {\n",
+      "    \"MONDO\": \"0002039\",\n",
+      "    \"id\": \"MONDO:0002039\",\n",
+      "    \"name\": \"cognitive disorder\",\n",
+      "    \"type\": \"biolink:Disease\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.8357342064986439,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.0945588244442226,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10102298,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"HP:0001251_HP:0000360_0.8401321539277617_07111840915054395_10809198194248215_10099363\",\n",
+      "  \"subject\": {\n",
+      "    \"HP\": \"0001251\",\n",
+      "    \"id\": \"HP:0001251\",\n",
+      "    \"name\": \"Ataxia\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.7111840915054395,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.0809198194248215,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10099363,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"CHEBI:4031_HP:0000360_0.8401321539277617_08886902851131677_10571543050145833_10098612\",\n",
+      "  \"subject\": {\n",
+      "    \"CHEBI\": \"4031\",\n",
+      "    \"id\": \"CHEBI:4031\",\n",
+      "    \"name\": \"cyclosporine\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.8886902851131677,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.0571543050145833,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10098612,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"MONDO:0001119_HP:0000360_0.8401321539277617_09257571836006121_1042593529879973_10100198\",\n",
+      "  \"subject\": {\n",
+      "    \"MONDO\": \"0001119\",\n",
+      "    \"id\": \"MONDO:0001119\",\n",
+      "    \"name\": \"premature menopause\",\n",
+      "    \"type\": \"biolink:Disease\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.9257571836006121,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.042593529879973,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10100198,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"MONDO:0013600_HP:0000360_0.8401321539277617_00_1040152790331654_10101431\",\n",
+      "  \"subject\": {\n",
+      "    \"MONDO\": \"0013600\",\n",
+      "    \"id\": \"MONDO:0013600\",\n",
+      "    \"name\": \"insomnia (disease)\",\n",
+      "    \"type\": \"biolink:Disease\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.0,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.040152790331654,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10101431,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "{\n",
+      "  \"_id\": \"CHEBI:15407_HP:0000360_0.8401321539277617_0002238827241549002_10328062673251501_10092760\",\n",
+      "  \"subject\": {\n",
+      "    \"CHEBI\": \"15407\",\n",
+      "    \"id\": \"CHEBI:15407\",\n",
+      "    \"name\": \"ephedrine\",\n",
+      "    \"type\": \"biolink:ChemicalEntity\"\n",
+      "  },\n",
+      "  \"association\": {\n",
+      "    \"predicate\": \"associated_with_increased_likelihood_of\",\n",
+      "    \"edge_attributes\": [\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:has_supporting_study_result\",\n",
+      "        \"value\": \"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+      "        \"attributes\": [\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+      "            \"value\": \"STATO:0000149\",\n",
+      "            \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:update_date\",\n",
+      "            \"value\": \"2022-05-18\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:p_value\",\n",
+      "            \"value\": 0.002238827241549002,\n",
+      "            \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"STATO:0000209\",\n",
+      "            \"value\": 0.8401321539277617,\n",
+      "            \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+      "            \"value\": 1.0328062673251501,\n",
+      "            \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+      "            \"value\": \"age < 18 excluded\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+      "            \"value\": \"2020-2022 (prediction)\"\n",
+      "          },\n",
+      "          {\n",
+      "            \"attribute_type_id\": \"biolink:supporting_study_size\",\n",
+      "            \"value\": 10092760,\n",
+      "            \"description\": \"The total number of patients or participants within a sample population.\"\n",
+      "          }\n",
+      "        ]\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+      "        \"value\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+      "        \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"attribute_type_id\": \"biolink:supporting_data_source\",\n",
+      "        \"value\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"value_type_id\": \"biolink:InformationResource\",\n",
+      "        \"value_url\": \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+      "        \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\"\n",
+      "      }\n",
+      "    ]\n",
+      "  },\n",
+      "  \"object\": {\n",
+      "    \"HP\": \"0000360\",\n",
+      "    \"id\": \"HP:0000360\",\n",
+      "    \"name\": \"Tinnitus\",\n",
+      "    \"type\": \"biolink:PhenotypicFeature\"\n",
+      "  },\n",
+      "  \"source\": {\n",
+      "    \"edge_sources\": [\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+      "        \"resource_role\": \"primary_knowledge_source\",\n",
+      "        \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+      "      },\n",
+      "      {\n",
+      "        \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+      "        \"resource_role\": \"supporting_data_source\"\n",
+      "      }\n",
+      "    ]\n",
+      "  }\n",
+      "}\n",
+      "Document IDs appear to be unique\n",
+      "\n",
+      "\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "import pandas as pd\n",
+    "import json\n",
+    "import sys, os\n",
+    "import numpy as np\n",
+    "\n",
+    "def parse_ehr_risk(data_folder):\n",
+    "\n",
+    "    edges_filename = \"ehr_risk_edges_data_2022_06_01.csv\"\n",
+    "    nodes_filename = \"ehr_risk_nodes_data_2022_06_01.csv\"\n",
+    "\n",
+    "    nodes_filepath = os.path.join(data_folder, nodes_filename)\n",
+    "    edges_filepath = os.path.join(data_folder, edges_filename)\n",
+    "    nodes_data = pd.read_csv(nodes_filepath, sep = ',')\n",
+    "    edges_data = pd.read_csv(edges_filepath, sep = ',')\n",
+    "    \n",
+    "    # the nodes file has duplicate ids; fix in enclave in future\n",
+    "    nodes_data = nodes_data.drop_duplicates(subset='id', keep=\"first\")\n",
+    "    \n",
+    "    # biolink category biolink:ChemicalSubstance has been deprecated. Use biolink:ChemicalEntity instead\n",
+    "    nodes_data[\"category\"].mask(nodes_data[\"category\"] == \"biolink:ChemicalSubstance\", \"biolink:ChemicalEntity\" , inplace=True )\n",
+    "\n",
+    "    # we originally provided the # of patients with condition --> log + patient count, and # of patients without condition --> log - patient count\n",
+    "    # get the approximate total number of patients in the study and call it \"total_sample_size\"\n",
+    "    edges_data[\"num_patients_with_condition\"] = 10**(edges_data['log_positive_patient_count']) # convert log pos patient count to an actual # \n",
+    "    edges_data[\"num_patients_without_condition\"] = 10**(edges_data['log_negative_patient_count']) # convert log neg patient count to an actual #\n",
+    "    edges_data = edges_data.drop(['log_positive_patient_count', 'log_negative_patient_count'], axis=1)\n",
+    "    edges_data[\"total_sample_size\"] = edges_data[\"num_patients_with_condition\"] + edges_data[\"num_patients_without_condition\"]\n",
+    "    edges_data = edges_data.drop(['num_patients_with_condition', 'num_patients_without_condition'], axis=1)\n",
+    "    edges_data[\"total_sample_size\"] = np.random.poisson(edges_data[\"total_sample_size\"]) # add poisson noise injection \n",
+    "\n",
+    "#     # create confidence interval column by concatenating 'lower_confidence_bound'and 'upper_confidence_bound', then dropping those columns\n",
+    "#     edges_data['log_odds_ratio_95_confidence_interval'] = edges_data.apply(lambda row: [row['lower_confidence_bound'], row['upper_confidence_bound']], axis=1)\n",
+    "#     edges_data = edges_data.drop(['lower_confidence_bound', 'upper_confidence_bound'], axis=1)\n",
+    "    \n",
+    "    #  ----- RE-CONSTRUCT KG FROM NODES AND EDGES FILES ------  #\n",
+    "    # merge the subject names, categories and ids from the nodes csv/table to the edges table\n",
+    "    kg = pd.merge(edges_data, nodes_data[['id', 'name', 'category']], left_on='subject', right_on = 'id', how=\"inner\")\n",
+    "    kg.rename(columns = {'category_x':'predicate_category',\n",
+    "                         'category_y': 'subject_category',\n",
+    "                         'id': 'subject_id',\n",
+    "                         'name': 'subject_name'}, inplace = True)\n",
+    "    # merge the object names, categories and ids from the nodes csv/table to the edges table\n",
+    "    kg = pd.merge(kg, nodes_data[['id', 'name', 'category']], left_on='object', right_on = 'id', how=\"inner\")\n",
+    "    kg.rename(columns = {'id':'object_id',\n",
+    "                     'category': 'object_category',\n",
+    "                     'name': 'object_name'}, inplace = True)\n",
+    "    #  ----- ------------------------------------------ ------  #\n",
+    "    \n",
+    "    # ensure there are no duplicates\n",
+    "    kg = kg.drop_duplicates(['subject', 'object', 'auc_roc', 'p_value', 'feature_coefficient'], keep='first')\n",
+    "    \n",
+    "    # some of the subjects/objects contain the string literal \"NONE\" (specific culprit is COVID Negative or something) Should look into this in future \n",
+    "    kg = kg[~kg[\"subject\"].str.contains(\"NONE\")==True]  # subject and object are all CURIEs, not names\n",
+    "    kg = kg[~kg[\"object\"].str.contains(\"NONE\")==True]\n",
+    "    kg = kg[~kg[\"subject\"].str.contains(\"none\")==True]\n",
+    "    kg = kg[~kg[\"object\"].str.contains(\"none\")==True]\n",
+    "    kg = kg[~kg[\"subject\"].str.contains(\"None\")==True]\n",
+    "    kg = kg[~kg[\"object\"].str.contains(\"None\")==True]\n",
+    "    \n",
+    "    id_list = [] # use this to check if your document IDs are unique. Collect them and see if they're all unique\n",
+    "    \n",
+    "    # iterate through each row in KG to yield json formatted triple\n",
+    "    for index, row in kg[:40].iterrows(): # comment for testing  \n",
+    "        id_dict = {} # this is the outter dict that holds inner dicts: subject_dict, association_dict, object_dict, and source_dict\n",
+    "        subject_dict = {} # inner dict\n",
+    "        association_dict = {} # inner dict\n",
+    "        object_dict = {} # inner dict\n",
+    "        source_dict = {} # inner dict (provides provenance as per TRAPI 1.4 standards)\n",
+    "\n",
+    "        # id generated by concatenating the following: subject_id CURIE, object_id CURIE, AUCROC (removing decimal point) and p-value (removing decimal point), feature coeffcient (removing decimal point), and total sample size\n",
+    "        doc_id = \"{}_{}_{}_{}_{}_{}\".format(row[\"subject\"],\n",
+    "                                       row[\"object\"],\n",
+    "                                       str(row['auc_roc']),\n",
+    "                                       str(row['p_value']).replace('.', ''),\n",
+    "                                       str(row['feature_coefficient']).replace('.', ''),\n",
+    "                                       str(row[\"total_sample_size\"]))\n",
+    "\n",
+    "        id_list.append(doc_id)\n",
+    "        id_dict[\"_id\"] = doc_id\n",
+    "        subject_dict[\"{}\".format(row[\"subject\"].split(':')[0])] = \"{}\".format(row[\"subject\"].split(':')[1]) # create the subject dict from the rows of the df \n",
+    "        subject_dict[\"id\"] = row[\"subject\"]\n",
+    "        subject_dict[\"name\"] = row[\"subject_name\"]\n",
+    "        subject_dict[\"type\"] = row[\"subject_category\"]\n",
+    "\n",
+    "        association_dict[\"predicate\"] = \"{}\".format(row[\"predicate\"].split(':')[1]) # create the association dict from the rows of the df. Edge attributes need extra work. The predicate is separated out into qualified predicate by X-BTE annotation, so we don't have to worry about qualifiers here\n",
+    "        association_dict[\"edge_attributes\"] = []\n",
+    "\n",
+    "        source_dict[\"edge_sources\"] = []\n",
+    "\n",
+    "        association_dict[\"edge_attributes\"].append(\n",
+    "            {\n",
+    "                \"attribute_type_id\":\"biolink:has_supporting_study_result\",\n",
+    "                \"value\":\"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+    "                \"attributes\": [\n",
+    "                    {\n",
+    "                        \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+    "                        \"value\": \"STATO:0000149\",\n",
+    "                        \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+    "                    },\n",
+    "                    {\n",
+    "                        \"attribute_type_id\":\"biolink:update_date\",\n",
+    "                        \"value\":row[\"provided_date\"]\n",
+    "                    },\n",
+    "                    {\n",
+    "                        \"attribute_type_id\": \"biolink:p_value\",\n",
+    "                        \"value\": row[\"p_value\"],\n",
+    "                        \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+    "                    },\n",
+    "                    {\n",
+    "                        \"attribute_type_id\": \"STATO:0000209\",\n",
+    "                        \"value\": row[\"auc_roc\"],\n",
+    "                        \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+    "                    },\n",
+    "                    {\n",
+    "                        \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+    "                        \"value\": row['feature_coefficient'],\n",
+    "                        \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+    "                    },\n",
+    "#                     {\n",
+    "#                         \"attribute_type_id\": \"biolink:log_odds_ratio_95_confidence_interval\",\n",
+    "#                         \"value\": row['log_odds_ratio_95_confidence_interval'],\n",
+    "#                         \"description\": \"log_odds_ratio_95_confidence_interval\"\n",
+    "#                     },\n",
+    "                    {\n",
+    "                        \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+    "                        \"value\": \"age < 18 excluded\"\n",
+    "                    },\n",
+    "                    {\n",
+    "                        \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+    "                        \"value\": \"2020-2022 (prediction)\"\n",
+    "                    },\n",
+    "                    {\n",
+    "                        \"attribute_type_id\": \"biolink:total_sample_size\",\n",
+    "                        \"value\": row[\"total_sample_size\"],\n",
+    "                        \"description\": \"The total number of patients or participants within a sample population.\"\n",
+    "                    }\n",
+    "                ]\n",
+    "            }\n",
+    "        )\n",
+    "        association_dict[\"edge_attributes\"].append(\n",
+    "            {\n",
+    "                \"attribute_type_id\":\"biolink:primary_knowledge_source\",\n",
+    "                \"value\":\"infores:biothings-multiomics-ehr-risk\",\n",
+    "                \"value_type_id\": \"biolink:InformationResource\",\n",
+    "                \"value_url\":  \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+    "                \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+    "            }\n",
+    "        )\n",
+    "        association_dict[\"edge_attributes\"].append(\n",
+    "            {\n",
+    "                \"attribute_type_id\":\"biolink:supporting_data_source\",\n",
+    "                \"value\":\"infores:providence-st-joseph-ehr\",\n",
+    "                \"value_type_id\": \"biolink:InformationResource\",\n",
+    "                \"value_url\":  \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+    "                \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\",\n",
+    "            }\n",
+    "        )\n",
+    "\n",
+    "        object_dict[\"{}\".format(row[\"object\"].split(':')[0])] = \"{}\".format(row[\"object\"].split(':')[1]) # create the object dict from the rows of the df \n",
+    "        object_dict[\"id\"] = row[\"object\"]\n",
+    "        object_dict[\"name\"] = row[\"object_name\"]\n",
+    "        object_dict[\"type\"] = row[\"object_category\"]\n",
+    "\n",
+    "        source_dict[\"edge_sources\"].append(\n",
+    "            {\n",
+    "                \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+    "                \"resource_role\": \"primary_knowledge_source\",\n",
+    "                \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+    "            }\n",
+    "        )\n",
+    "\n",
+    "        source_dict[\"edge_sources\"].append(\n",
+    "            {\n",
+    "                \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+    "                \"resource_role\": \"supporting_data_source\"\n",
+    "            }\n",
+    "        )\n",
+    "\n",
+    "        id_dict[\"subject\"] = subject_dict # put the subject, association, object, and source dicts into the outer dict called id_dict\n",
+    "        id_dict[\"association\"] = association_dict\n",
+    "        id_dict[\"object\"] = object_dict\n",
+    "        id_dict[\"source\"] = source_dict\n",
+    "            \n",
+    "        # throw error for any rows that are missing any relevant values, such as subject name, subject id/CURIE, subject category, p-value, etc...\n",
+    "        try:\n",
+    "            assert not {x for x in {row[\"total_sample_size\"],\n",
+    "                                    row[\"subject\"],\n",
+    "                                    row[\"subject_name\"],\n",
+    "                                    row[\"subject_category\"],\n",
+    "                                    row[\"object\"],\n",
+    "                                    row[\"object_name\"],\n",
+    "                                    row[\"object_category\"],\n",
+    "                                    row[\"p_value\"],\n",
+    "                                    row[\"auc_roc\"],\n",
+    "                                    row['feature_coefficient']} if x in {None,\n",
+    "                                                                         \"NONE\",\n",
+    "                                                                         \"None\",\n",
+    "                                                                         \"none\",\n",
+    "                                                                         \"NA\"}}, \"Error: All values including subject and object IDs, categories, names, p-value, AUC-ROC, and feature coefficient must be non-null and not contain string literal None or NONE\"\n",
+    "            print(json.dumps(id_dict, indent=2)) # uncomment for testing\n",
+    "#             print(index) # uncomment for testing\n",
+    "#             yield id_dict # comment for testing\n",
+    "        except AssertionError as msg:\n",
+    "            print(msg)\n",
+    "\n",
+    "    if len(id_list) != len(set(id_list)):\n",
+    "        print(\"You do not have unique document IDs for each edge in your KG. Either you have duplicate rows/edges, or you simply didn't make a unique identifer (Document ID) for each one.\\n\\n\\n\")\n",
+    "    else:\n",
+    "        print(\"Document IDs appear to be unique\\n\\n\\n\")\n",
+    "        \n",
+    "        \n",
+    "    "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "f3a074a8",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import pandas as pd\n",
+    "import json\n",
+    "import sys, os\n",
+    "import numpy as np\n",
+    "\n",
+    "def parse_ehr_risk(data_folder):\n",
+    "\n",
+    "    edges_filename = \"ehr_risk_edges_data_2022_06_01.csv\"\n",
+    "    nodes_filename = \"ehr_risk_nodes_data_2022_06_01.csv\"\n",
+    "\n",
+    "    nodes_filepath = os.path.join(data_folder, nodes_filename)\n",
+    "    edges_filepath = os.path.join(data_folder, edges_filename)\n",
+    "    nodes_data = pd.read_csv(nodes_filepath, sep = ',')\n",
+    "    edges_data = pd.read_csv(edges_filepath, sep = ',')\n",
+    "    \n",
+    "    # the nodes file has duplicate ids; fix in enclave in future\n",
+    "    nodes_data = nodes_data.drop_duplicates(subset='id', keep=\"first\")\n",
+    "    \n",
+    "    # biolink category biolink:ChemicalSubstance has been deprecated. Use biolink:ChemicalEntity instead\n",
+    "    nodes_data[\"category\"].mask(nodes_data[\"category\"] == \"biolink:ChemicalSubstance\", \"biolink:ChemicalEntity\" , inplace=True )\n",
+    "\n",
+    "    # we originally provided the # of patients with condition --> log + patient count, and # of patients without condition --> log - patient count\n",
+    "    # get the approximate total number of patients in the study and call it \"total_sample_size\"\n",
+    "    edges_data[\"num_patients_with_condition\"] = 10**(edges_data['log_positive_patient_count']) # convert log pos patient count to an actual # \n",
+    "    edges_data[\"num_patients_without_condition\"] = 10**(edges_data['log_negative_patient_count']) # convert log neg patient count to an actual #\n",
+    "    edges_data = edges_data.drop(['log_positive_patient_count', 'log_negative_patient_count'], axis=1)\n",
+    "    edges_data[\"total_sample_size\"] = edges_data[\"num_patients_with_condition\"] + edges_data[\"num_patients_without_condition\"]\n",
+    "    edges_data = edges_data.drop(['num_patients_with_condition', 'num_patients_without_condition'], axis=1)\n",
+    "    edges_data[\"total_sample_size\"] = np.random.poisson(edges_data[\"total_sample_size\"]) # add poisson noise injection \n",
+    "\n",
+    "#     # create confidence interval column by concatenating 'lower_confidence_bound'and 'upper_confidence_bound', then dropping those columns\n",
+    "#     edges_data['log_odds_ratio_95_confidence_interval'] = edges_data.apply(lambda row: [row['lower_confidence_bound'], row['upper_confidence_bound']], axis=1)\n",
+    "#     edges_data = edges_data.drop(['lower_confidence_bound', 'upper_confidence_bound'], axis=1)\n",
+    "    \n",
+    "    #  ----- RE-CONSTRUCT KG FROM NODES AND EDGES FILES ------  #\n",
+    "    # merge the subject names, categories and ids from the nodes csv/table to the edges table\n",
+    "    kg = pd.merge(edges_data, nodes_data[['id', 'name', 'category']], left_on='subject', right_on = 'id', how=\"inner\")\n",
+    "    kg.rename(columns = {'category_x':'predicate_category',\n",
+    "                         'category_y': 'subject_category',\n",
+    "                         'id': 'subject_id',\n",
+    "                         'name': 'subject_name'}, inplace = True)\n",
+    "    # merge the object names, categories and ids from the nodes csv/table to the edges table\n",
+    "    kg = pd.merge(kg, nodes_data[['id', 'name', 'category']], left_on='object', right_on = 'id', how=\"inner\")\n",
+    "    kg.rename(columns = {'id':'object_id',\n",
+    "                     'category': 'object_category',\n",
+    "                     'name': 'object_name'}, inplace = True)\n",
+    "    #  ----- ------------------------------------------ ------  #\n",
+    "    \n",
+    "    # ensure there are no duplicates\n",
+    "    kg = kg.drop_duplicates(['subject', 'object', 'auc_roc', 'p_value', 'feature_coefficient'], keep='first')\n",
+    "    \n",
+    "    # some of the subjects/objects contain the string literal \"NONE\" (specific culprit is COVID Negative or something) Should look into this in future \n",
+    "    kg = kg[~kg[\"subject\"].str.contains(\"NONE\")==True]  # subject and object are all CURIEs, not names\n",
+    "    kg = kg[~kg[\"object\"].str.contains(\"NONE\")==True]\n",
+    "    kg = kg[~kg[\"subject\"].str.contains(\"none\")==True]\n",
+    "    kg = kg[~kg[\"object\"].str.contains(\"none\")==True]\n",
+    "    kg = kg[~kg[\"subject\"].str.contains(\"None\")==True]\n",
+    "    kg = kg[~kg[\"object\"].str.contains(\"None\")==True]\n",
+    "    \n",
+    "    id_list = [] # use this to check if your document IDs are unique. Collect them and see if they're all unique\n",
+    "    \n",
+    "    # iterate through each row in KG to yield json formatted triple\n",
+    "    for index, row in kg[:40].iterrows(): # comment for testing  \n",
+    "        id_dict = {} # this is the outter dict that holds inner dicts: subject_dict, association_dict, object_dict, and source_dict\n",
+    "        subject_dict = {} # inner dict\n",
+    "        association_dict = {} # inner dict\n",
+    "        object_dict = {} # inner dict\n",
+    "        source_dict = {} # inner dict (provides provenance as per TRAPI 1.4 standards)\n",
+    "\n",
+    "        # id generated by concatenating the following: subject_id CURIE, object_id CURIE, AUCROC (removing decimal point) and p-value (removing decimal point), feature coeffcient (removing decimal point), and total sample size\n",
+    "        doc_id = \"{}_{}_{}_{}_{}_{}\".format(row[\"subject\"],\n",
+    "                                       row[\"object\"],\n",
+    "                                       str(row['auc_roc']),\n",
+    "                                       str(row['p_value']).replace('.', ''),\n",
+    "                                       str(row['feature_coefficient']).replace('.', ''),\n",
+    "                                       str(row[\"total_sample_size\"]))\n",
+    "\n",
+    "        id_list.append(doc_id)\n",
+    "        id_dict[\"_id\"] = doc_id\n",
+    "        subject_dict[\"{}\".format(row[\"subject\"].split(':')[0])] = \"{}\".format(row[\"subject\"].split(':')[1]) # create the subject dict from the rows of the df \n",
+    "        subject_dict[\"id\"] = row[\"subject\"]\n",
+    "        subject_dict[\"name\"] = row[\"subject_name\"]\n",
+    "        subject_dict[\"type\"] = row[\"subject_category\"]\n",
+    "\n",
+    "        association_dict[\"predicate\"] = \"{}\".format(row[\"predicate\"].split(':')[1]) # create the association dict from the rows of the df. Edge attributes need extra work. The predicate is separated out into qualified predicate by X-BTE annotation, so we don't have to worry about qualifiers here\n",
+    "        association_dict[\"edge_attributes\"] = []\n",
+    "\n",
+    "        source_dict[\"edge_sources\"] = []\n",
+    "\n",
+    "        association_dict[\"edge_attributes\"].append(\n",
+    "            {\n",
+    "                \"attribute_type_id\":\"biolink:has_supporting_study_result\",\n",
+    "                \"value\":\"We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+    "                \"attributes\": [\n",
+    "                    {\n",
+    "                        \"attribute_type_id\": \"biolink:supporting_study_method_type\",\n",
+    "                        \"value\": \"STATO:0000149\",\n",
+    "                        \"description\": \"Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)\"\n",
+    "                    },\n",
+    "                    {\n",
+    "                        \"attribute_type_id\":\"biolink:update_date\",\n",
+    "                        \"value\":row[\"provided_date\"]\n",
+    "                    },\n",
+    "                    {\n",
+    "                        \"attribute_type_id\": \"biolink:p_value\",\n",
+    "                        \"value\": row[\"p_value\"],\n",
+    "                        \"description\": \"The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.\"\n",
+    "                    },\n",
+    "                    {\n",
+    "                        \"attribute_type_id\": \"STATO:0000209\",\n",
+    "                        \"value\": row[\"auc_roc\"],\n",
+    "                        \"description\": \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"\n",
+    "                    },\n",
+    "                    {\n",
+    "                        \"attribute_type_id\": \"biolink:log_odds_ratio\",\n",
+    "                        \"value\": row['feature_coefficient'],\n",
+    "                        \"description\": \"The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.\"\n",
+    "                    },\n",
+    "#                     {\n",
+    "#                         \"attribute_type_id\": \"biolink:log_odds_ratio_95_confidence_interval\",\n",
+    "#                         \"value\": row['log_odds_ratio_95_confidence_interval'],\n",
+    "#                         \"description\": \"log_odds_ratio_95_confidence_interval\"\n",
+    "#                     },\n",
+    "                    {\n",
+    "                        \"attribute_type_id\": \"biolink:supporting_study_cohort\",\n",
+    "                        \"value\": \"age < 18 excluded\"\n",
+    "                    },\n",
+    "                    {\n",
+    "                        \"attribute_type_id\": \"biolink:supporting_study_date_range\",\n",
+    "                        \"value\": \"2020-2022 (prediction)\"\n",
+    "                    },\n",
+    "                    {\n",
+    "                        \"attribute_type_id\": \"biolink:total_sample_size\",\n",
+    "                        \"value\": row[\"total_sample_size\"],\n",
+    "                        \"description\": \"The total number of patients or participants within a sample population.\"\n",
+    "                    }\n",
+    "                ]\n",
+    "            }\n",
+    "        )\n",
+    "        association_dict[\"edge_attributes\"].append(\n",
+    "            {\n",
+    "                \"attribute_type_id\":\"biolink:primary_knowledge_source\",\n",
+    "                \"value\":\"infores:biothings-multiomics-ehr-risk\",\n",
+    "                \"value_type_id\": \"biolink:InformationResource\",\n",
+    "                \"value_url\":  \"http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a\",\n",
+    "                \"description\": \"The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).\",\n",
+    "            }\n",
+    "        )\n",
+    "        association_dict[\"edge_attributes\"].append(\n",
+    "            {\n",
+    "                \"attribute_type_id\":\"biolink:supporting_data_source\",\n",
+    "                \"value\":\"infores:providence-st-joseph-ehr\",\n",
+    "                \"value_type_id\": \"biolink:InformationResource\",\n",
+    "                \"value_url\":  \"https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP\",\n",
+    "                \"description\": \"A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.\",\n",
+    "            }\n",
+    "        )\n",
+    "\n",
+    "        object_dict[\"{}\".format(row[\"object\"].split(':')[0])] = \"{}\".format(row[\"object\"].split(':')[1]) # create the object dict from the rows of the df \n",
+    "        object_dict[\"id\"] = row[\"object\"]\n",
+    "        object_dict[\"name\"] = row[\"object_name\"]\n",
+    "        object_dict[\"type\"] = row[\"object_category\"]\n",
+    "\n",
+    "        source_dict[\"edge_sources\"].append(\n",
+    "            {\n",
+    "                \"resource_id\": \"infores:biothings-multiomics-ehr-risk\",\n",
+    "                \"resource_role\": \"primary_knowledge_source\",\n",
+    "                \"upstream_resource_ids\": \"infores:providence-st-joseph-ehr\"\n",
+    "            }\n",
+    "        )\n",
+    "\n",
+    "        source_dict[\"edge_sources\"].append(\n",
+    "            {\n",
+    "                \"resource_id\": \"infores:providence-st-joseph-ehr\",\n",
+    "                \"resource_role\": \"supporting_data_source\"\n",
+    "            }\n",
+    "        )\n",
+    "\n",
+    "        id_dict[\"subject\"] = subject_dict # put the subject, association, object, and source dicts into the outer dict called id_dict\n",
+    "        id_dict[\"association\"] = association_dict\n",
+    "        id_dict[\"object\"] = object_dict\n",
+    "        id_dict[\"source\"] = source_dict\n",
+    "            \n",
+    "        # throw error for any rows that are missing any relevant values, such as subject name, subject id/CURIE, subject category, p-value, etc...\n",
+    "\n",
+    "        try:\n",
+    "            assert not {x for x in {row[\"total_sample_size\"],\n",
+    "                                    row[\"subject\"],\n",
+    "                                    row[\"subject_name\"],\n",
+    "                                    row[\"subject_category\"],\n",
+    "                                    row[\"object\"],\n",
+    "                                    row[\"object_name\"],\n",
+    "                                    row[\"object_category\"],\n",
+    "                                    row[\"p_value\"],\n",
+    "                                    row[\"auc_roc\"],\n",
+    "                                    row['feature_coefficient']} if x in {None,\n",
+    "                                                                         \"NONE\",\n",
+    "                                                                         \"None\",\n",
+    "                                                                         \"none\",\n",
+    "                                                                         \"NA\"}}, \"Error: All values including subject and object IDs, categories, names, p-value, AUC-ROC, and feature coefficient must be non-null and not contain string literal None or NONE\"\n",
+    "            print(json.dumps(id_dict, indent=2)) # uncomment for testing\n",
+    "#             print(index) # uncomment for testing\n",
+    "#             yield id_dict # comment for testing\n",
+    "        except AssertionError as msg:\n",
+    "            print(msg)\n",
+    "\n",
+    "    if len(id_list) != len(set(id_list)):\n",
+    "        print(\"You do not have unique document IDs for each edge in your KG. Either you have duplicate rows/edges, or you simply didn't make a unique identifer (Document ID) for each one.\\n\\n\\n\")\n",
+    "    else:\n",
+    "        print(\"Document IDs appear to be unique\\n\\n\\n\")\n",
+    "\n",
+    "\n",
+    "\n",
+    "\n",
+    "    "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 151,
+   "id": "03180ac7",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Document IDs appear to be unique\n",
+      "\n",
+      "[{'_id': 'HP:0008629_HP:0000360_0.8401321539277617_00_8796399245685702_10096539', 'subject': {'HP': '0008629', 'id': 'HP:0008629', 'name': 'Pulsatile tinnitus', 'type': 'biolink:PhenotypicFeature'}, 'association': {'predicate': 'associated_with_increased_likelihood_of', 'edge_attributes': [{'attribute_type_id': 'biolink:has_supporting_study_result', 'value': 'We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).', 'attributes': [{'attribute_type_id': 'biolink:supporting_study_method_type', 'value': 'STATO:0000149', 'description': 'Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)'}, {'attribute_type_id': 'biolink:update_date', 'value': '2022-05-18'}, {'attribute_type_id': 'biolink:p_value', 'value': 0.0, 'description': 'The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.'}, {'attribute_type_id': 'STATO:0000209', 'value': 0.8401321539277617, 'description': \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"}, {'attribute_type_id': 'biolink:log_odds_ratio', 'value': 8.796399245685702, 'description': 'The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.'}, {'attribute_type_id': 'biolink:supporting_study_cohort', 'value': 'age < 18 excluded'}, {'attribute_type_id': 'biolink:supporting_study_date_range', 'value': '2020-2022 (prediction)'}, {'attribute_type_id': 'biolink:supporting_study_size', 'value': 10096539, 'description': 'The total number of patients or participants within a sample population.'}]}, {'attribute_type_id': 'biolink:primary_knowledge_source', 'value': 'infores:biothings-multiomics-ehr-risk', 'value_type_id': 'biolink:InformationResource', 'value_url': 'http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a', 'description': 'The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).'}, {'attribute_type_id': 'biolink:supporting_data_source', 'value': 'infores:providence-st-joseph-ehr', 'value_type_id': 'biolink:InformationResource', 'value_url': 'https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP', 'description': 'A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.'}]}, 'object': {'HP': '0000360', 'id': 'HP:0000360', 'name': 'Tinnitus', 'type': 'biolink:PhenotypicFeature'}, 'source': {'edge_sources': [{'resource_id': 'infores:biothings-multiomics-ehr-risk', 'resource_role': 'primary_knowledge_source', 'upstream_resource_ids': 'infores:providence-st-joseph-ehr'}, {'resource_id': 'infores:providence-st-joseph-ehr', 'resource_role': 'supporting_data_source'}]}}, {'_id': 'MONDO:0010643_HP:0000360_0.8401321539277617_09998721067797812_8585212287149526_10100701', 'subject': {'MONDO': '0010643', 'id': 'MONDO:0010643', 'name': 'acute leukemia (disease)', 'type': 'biolink:Disease'}, 'association': {'predicate': 'associated_with_increased_likelihood_of', 'edge_attributes': [{'attribute_type_id': 'biolink:has_supporting_study_result', 'value': 'We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).', 'attributes': [{'attribute_type_id': 'biolink:supporting_study_method_type', 'value': 'STATO:0000149', 'description': 'Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)'}, {'attribute_type_id': 'biolink:update_date', 'value': '2022-05-18'}, {'attribute_type_id': 'biolink:p_value', 'value': 0.9998721067797812, 'description': 'The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.'}, {'attribute_type_id': 'STATO:0000209', 'value': 0.8401321539277617, 'description': \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"}, {'attribute_type_id': 'biolink:log_odds_ratio', 'value': 8.585212287149526, 'description': 'The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.'}, {'attribute_type_id': 'biolink:supporting_study_cohort', 'value': 'age < 18 excluded'}, {'attribute_type_id': 'biolink:supporting_study_date_range', 'value': '2020-2022 (prediction)'}, {'attribute_type_id': 'biolink:supporting_study_size', 'value': 10100701, 'description': 'The total number of patients or participants within a sample population.'}]}, {'attribute_type_id': 'biolink:primary_knowledge_source', 'value': 'infores:biothings-multiomics-ehr-risk', 'value_type_id': 'biolink:InformationResource', 'value_url': 'http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a', 'description': 'The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).'}, {'attribute_type_id': 'biolink:supporting_data_source', 'value': 'infores:providence-st-joseph-ehr', 'value_type_id': 'biolink:InformationResource', 'value_url': 'https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP', 'description': 'A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.'}]}, 'object': {'HP': '0000360', 'id': 'HP:0000360', 'name': 'Tinnitus', 'type': 'biolink:PhenotypicFeature'}, 'source': {'edge_sources': [{'resource_id': 'infores:biothings-multiomics-ehr-risk', 'resource_role': 'primary_knowledge_source', 'upstream_resource_ids': 'infores:providence-st-joseph-ehr'}, {'resource_id': 'infores:providence-st-joseph-ehr', 'resource_role': 'supporting_data_source'}]}}, {'_id': 'UNII:25ADE2236L_HP:0000360_0.8401321539277617_09367666401584368_4558176672832635_10098743', 'subject': {'UNII': '25ADE2236L', 'id': 'UNII:25ADE2236L', 'name': 'thrombin', 'type': 'biolink:ChemicalSubstance'}, 'association': {'predicate': 'associated_with_increased_likelihood_of', 'edge_attributes': [{'attribute_type_id': 'biolink:has_supporting_study_result', 'value': 'We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).', 'attributes': [{'attribute_type_id': 'biolink:supporting_study_method_type', 'value': 'STATO:0000149', 'description': 'Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)'}, {'attribute_type_id': 'biolink:update_date', 'value': '2022-05-18'}, {'attribute_type_id': 'biolink:p_value', 'value': 0.9367666401584368, 'description': 'The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.'}, {'attribute_type_id': 'STATO:0000209', 'value': 0.8401321539277617, 'description': \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"}, {'attribute_type_id': 'biolink:log_odds_ratio', 'value': 4.558176672832635, 'description': 'The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.'}, {'attribute_type_id': 'biolink:supporting_study_cohort', 'value': 'age < 18 excluded'}, {'attribute_type_id': 'biolink:supporting_study_date_range', 'value': '2020-2022 (prediction)'}, {'attribute_type_id': 'biolink:supporting_study_size', 'value': 10098743, 'description': 'The total number of patients or participants within a sample population.'}]}, {'attribute_type_id': 'biolink:primary_knowledge_source', 'value': 'infores:biothings-multiomics-ehr-risk', 'value_type_id': 'biolink:InformationResource', 'value_url': 'http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a', 'description': 'The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).'}, {'attribute_type_id': 'biolink:supporting_data_source', 'value': 'infores:providence-st-joseph-ehr', 'value_type_id': 'biolink:InformationResource', 'value_url': 'https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP', 'description': 'A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.'}]}, 'object': {'HP': '0000360', 'id': 'HP:0000360', 'name': 'Tinnitus', 'type': 'biolink:PhenotypicFeature'}, 'source': {'edge_sources': [{'resource_id': 'infores:biothings-multiomics-ehr-risk', 'resource_role': 'primary_knowledge_source', 'upstream_resource_ids': 'infores:providence-st-joseph-ehr'}, {'resource_id': 'infores:providence-st-joseph-ehr', 'resource_role': 'supporting_data_source'}]}}, {'_id': 'UNII:K16AIQ8CTM_HP:0000360_0.8401321539277617_09985626800193924_43575215395209606_10097768', 'subject': {'UNII': 'K16AIQ8CTM', 'id': 'UNII:K16AIQ8CTM', 'name': 'pertuzumab', 'type': 'biolink:ChemicalSubstance'}, 'association': {'predicate': 'associated_with_increased_likelihood_of', 'edge_attributes': [{'attribute_type_id': 'biolink:has_supporting_study_result', 'value': 'We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).', 'attributes': [{'attribute_type_id': 'biolink:supporting_study_method_type', 'value': 'STATO:0000149', 'description': 'Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)'}, {'attribute_type_id': 'biolink:update_date', 'value': '2022-05-18'}, {'attribute_type_id': 'biolink:p_value', 'value': 0.9985626800193924, 'description': 'The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.'}, {'attribute_type_id': 'STATO:0000209', 'value': 0.8401321539277617, 'description': \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"}, {'attribute_type_id': 'biolink:log_odds_ratio', 'value': 4.3575215395209606, 'description': 'The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.'}, {'attribute_type_id': 'biolink:supporting_study_cohort', 'value': 'age < 18 excluded'}, {'attribute_type_id': 'biolink:supporting_study_date_range', 'value': '2020-2022 (prediction)'}, {'attribute_type_id': 'biolink:supporting_study_size', 'value': 10097768, 'description': 'The total number of patients or participants within a sample population.'}]}, {'attribute_type_id': 'biolink:primary_knowledge_source', 'value': 'infores:biothings-multiomics-ehr-risk', 'value_type_id': 'biolink:InformationResource', 'value_url': 'http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a', 'description': 'The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).'}, {'attribute_type_id': 'biolink:supporting_data_source', 'value': 'infores:providence-st-joseph-ehr', 'value_type_id': 'biolink:InformationResource', 'value_url': 'https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP', 'description': 'A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.'}]}, 'object': {'HP': '0000360', 'id': 'HP:0000360', 'name': 'Tinnitus', 'type': 'biolink:PhenotypicFeature'}, 'source': {'edge_sources': [{'resource_id': 'infores:biothings-multiomics-ehr-risk', 'resource_role': 'primary_knowledge_source', 'upstream_resource_ids': 'infores:providence-st-joseph-ehr'}, {'resource_id': 'infores:providence-st-joseph-ehr', 'resource_role': 'supporting_data_source'}]}}, {'_id': 'MONDO:0007972_HP:0000360_0.8401321539277617_009395878968875304_392606416950393_10102280', 'subject': {'MONDO': '0007972', 'id': 'MONDO:0007972', 'name': 'Meniere disease', 'type': 'biolink:Disease'}, 'association': {'predicate': 'associated_with_increased_likelihood_of', 'edge_attributes': [{'attribute_type_id': 'biolink:has_supporting_study_result', 'value': 'We train a large collection of multivariable, binary logistic regression models on EHR data for each specific condition/disease/outcome. Features include labs, medications, and phenotypes. Directed edges point from risk factors to specific outcomes (diseases, phenotype, or medication exposure).', 'attributes': [{'attribute_type_id': 'biolink:supporting_study_method_type', 'value': 'STATO:0000149', 'description': 'Binomial logistic regression for analysis of dichotomous dependent variable (in this case, for having this particular condition/disease/outcome or not)'}, {'attribute_type_id': 'biolink:update_date', 'value': '2022-05-18'}, {'attribute_type_id': 'biolink:p_value', 'value': 0.09395878968875304, 'description': 'The p-value represents the probability of observing the estimated coefficient (or more extreme value) under the assumption of the null hypothesis (which assumes that there is no relationship between the independent variable and outcome variable). The p-value associated with each coefficient helps determine whether the relationship between the independent variable and the outcome is statistically significant. A low p-value suggests that the observed relationship between the independent variable and the outcome is unlikely to occur by chance alone, providing evidence against the null hypothesis.'}, {'attribute_type_id': 'STATO:0000209', 'value': 0.8401321539277617, 'description': \"The AUROC provides a way to evaluate the model's ability to discriminate between the two classes (the presenece of absence of condition/disease/outcome). Values range between 0-1; the higher the AUROC, the better the model's ability to discriminate between clasess.\"}, {'attribute_type_id': 'biolink:log_odds_ratio', 'value': 3.92606416950393, 'description': 'The logarithm of the odds ratio (log odds ratio), or the ratio of the odds of event Y occurring in an exposed group versus the odds of event Y occurring in a non-exposed group.'}, {'attribute_type_id': 'biolink:supporting_study_cohort', 'value': 'age < 18 excluded'}, {'attribute_type_id': 'biolink:supporting_study_date_range', 'value': '2020-2022 (prediction)'}, {'attribute_type_id': 'biolink:supporting_study_size', 'value': 10102280, 'description': 'The total number of patients or participants within a sample population.'}]}, {'attribute_type_id': 'biolink:primary_knowledge_source', 'value': 'infores:biothings-multiomics-ehr-risk', 'value_type_id': 'biolink:InformationResource', 'value_url': 'http://smart-api.info/registry?q=d86a24f6027ffe778f84ba10a7a1861a', 'description': 'The EHR Risk KP is created and maintained by the Multiomics Provider team from the Institute for Systems Biology in Seattle, WA. Through a partnership with Providence/Swedish Health Services and Institute for Systems Biology, we analyze over 26 million EHRs. We use these records to train a large collection of interpretable machine learning models which are integrated into a single large Knowledge Graph, with directed edges pointing from risk factors to specific outcomes (diseases, phenotype, or medication exposure).'}, {'attribute_type_id': 'biolink:supporting_data_source', 'value': 'infores:providence-st-joseph-ehr', 'value_type_id': 'biolink:InformationResource', 'value_url': 'https://github.com/NCATSTranslator/Translator-All/wiki/EHR-Risk-KP', 'description': 'A partnership with Providence/Swedish Health Services and Institute for Systems Biology allows analysis of 26 million EHRs from patients in seven states in the US, including Alaska, California, Montana, Oregon, Washington, Texas, and New Mexico. Please email data-access@isbscience.org for more information.'}]}, 'object': {'HP': '0000360', 'id': 'HP:0000360', 'name': 'Tinnitus', 'type': 'biolink:PhenotypicFeature'}, 'source': {'edge_sources': [{'resource_id': 'infores:biothings-multiomics-ehr-risk', 'resource_role': 'primary_knowledge_source', 'upstream_resource_ids': 'infores:providence-st-joseph-ehr'}, {'resource_id': 'infores:providence-st-joseph-ehr', 'resource_role': 'supporting_data_source'}]}}]\n"
+     ]
+    }
+   ],
+   "source": [
+    "def main():\n",
+    "    data_folder = \"../../data\" # uncomment for testing\n",
+    "    parse_ehr_risk(data_folder) \n",
+    "\n",
+    "    \n",
+    "if __name__ == \"__main__\":\n",
+    "    main()\n",
+    "\n",
+    "# def f(): return list(parse_ehr_risk(data_folder))\n",
+    "\n",
+    "# print(f())"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "4f8e6295",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "b391313d",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "a9131832",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "7d579a7c",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "3deaf666",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "38af4ce6",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "975b39b8",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "ee8cb830",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "e8da035e",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "36dbb201",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "ea4aaa3b",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "1993c430",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "5c1a5344",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "d2b7d630",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "87af529d",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "4d0a883a",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "855f14b3",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "b9dd7dd4",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "dffef3dd",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "a1f1b174",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "14117bf5",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "5b4f2f35",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "f620ac0c",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "f9f8ddcb",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "2c12970b",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "6075e928",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "b885d179",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "0ad7fd3a",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "31661afb",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "33b5bdf8",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "5f2de47d",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "fc5ba61b",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "dfbc4354",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "6ed22495",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "0005db9e",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "04e71223",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "76b3e63f",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "6399bef2",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "508fcd23",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "80e6c8da",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "81f51f0b",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "ddcf59bd",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "d2f7626b",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "43f24102",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "6b054438",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "b600cab2",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "a26f5991",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "d19cabbf",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "ed293615",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "fcff8936",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "67cea542",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "4683b850",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "53da2865",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "### from UI, we found duplicate predicate in June 6 2020 edges files"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "658cef8a",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "edges_data.loc[(edges_data['subject'] == 'MONDO:0011849') & (edges_data['object'] == 'MONDO:0005083')]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "61b5337d",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "edges_data.loc[(edges_data['subject'] == 'MONDO:0005083') & (edges_data['object'] == 'MONDO:0011849')]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "c149a94c",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "https://github.com/uhbrar/ReasonerAPI/blob/update_guide/MigrationAndImplementationGuide1-4.md"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "ed91434c",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "variable names to use for Clinical Data Committee:\n",
+    "log_odds_ratio, total_sample_size, log_odds_ratio_95_confidence_interval"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "cecd7f59",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# example 1.4 output from molepro\n",
+    "{\n",
+    "  \t\"attribute_type_id\": \"biolink:aggregator_knowledge_source\",\t\n",
+    "  \t\"value\": \"infores:molepro\",      \n",
+    "  \t\"value_type_id\": \"biolink:InformationResource\",    \n",
+    "  \t\"value_url\":  \"https://translator.broadinstitute.org/molepro/trapi/v1.0\",\n",
+    "  \t\"description\": \"The Molecular Data Provider KP from NCATS Translator\",\n",
+    "  \t\"attribute_source\": \"infores:molepro\"\n",
+    "       },\n",
+    "\t{\n",
+    "  \t\"attribute_type_id\": \"biolink:aggregator_knowledge_source\",\n",
+    "  \t\"value\":  \"infores:chembl\",\n",
+    "  \t\"value_type_id\": \"biolink:InformationResource\",    \n",
+    "  \t\"value_url\":  \"https://www.ebi.ac.uk/chembl\",\n",
+    "  \t\"description\": \"ChEMBL is a manually curated database of bioactive molecules...\",\n",
+    "  \t\"attribute_source\": \"infores:molepro\"\n",
+    "\t},\n",
+    "\t{\n",
+    "  \t\"attribute_type_id\": \"biolink:primary_knowledge_source\",\n",
+    "  \t\"value\":  \"infores:clinical-trials-gov\", \n",
+    "  \t\"value_type_id\": \"biolink:InformationResource\",    \n",
+    "  \t\"value_url\":  \"https://www.clinicaltrials.gov\",\n",
+    "  \t\"description\": \"ClinicalTrials.gov is...\",\n",
+    "  \t\"attribute_source\": \"infores:chembl\"\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "c3a5ca8d",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "\"edges\": [\n",
+    "  {\n",
+    "  \"id\": \"Association002\",   \n",
+    "  \"category\": \"biolink:FeatureVariableAssociation\",\n",
+    "  \"subject\": \"ncit:C29886\" # Airborne Particulate Matter (PM2.5),   \t \n",
+    "  \"predicate\": \"biolink:correlates_with\",\n",
+    "  \"object\": \"tvfo:xxxxx\" # t.b.d. term for 'ED Visits for Asthma',\n",
+    "  \"subject_modifier\": \"biolink:Exposure\",\n",
+    "  \"attributes\":  [\n",
+    "\t{\n",
+    "  \t\"attribute_type_id\": \"biolink:primary_knowledge_source\",  \n",
+    "  \t\"value\":  \"infores:icees-asthma\",  \n",
+    "  \t\"value_type_id\": \"biolink:InformationResource\",\t \n",
+    "  \t\"value_url\":  \"https://icees.renci.org:16339\",\n",
+    "  \t\"description\": \"The ICEES Provider ...\",\n",
+    "  \t\"attribute_source\": \"infores:icees-asthma\"\n",
+    "\t},\n",
+    "\t{\n",
+    "  \t\"attribute_type_id\": \"biolink:supporting_data_source\",  \n",
+    "  \t\"value\":  \"infores:us-epa-airborne-pollutant-exposures-data\",\n",
+    "  \t\"value_type_id\": \"biolink:InformationResource\",\t \n",
+    "  \t\"description\": \"US Environmental Protection Agency Airborne Pollutant Exposure Data\",\n",
+    "  \t\"attribute_source\": \"infores:icees-asthma\"\n",
+    "\t},\n",
+    "\t{\n",
+    "  \t\"attribute_type_id\": \"biolink:supporting_data_source\",  \n",
+    "  \t\"value\":  \"infores:unc-cdw-health\",\n",
+    "  \t\"value_type_id\": \"biolink:InformationResource\",\t \n",
+    "  \t\"description\": \"UNC Carolina Data Warehouse for Health Patient EHR Data\",\n",
+    "  \t\"attribute_source\": \"infores:icees-asthma\"\n",
+    "\t}\n",
+    "     ]\n",
+    "  }\n",
+    "]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "df78b8ff",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "edges_sig[\"num_patients_with_condition\"] = 10**(edges_sig['log_positive_patient_count'])\n",
+    "edges_sig[\"num_patients_without_condition\"] = 10**(edges_sig['log_negative_patient_count'])\n",
+    "edges_sig"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.7.6"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}