Improve association table search

backend/src/monarch_py/cli.py

@@ -247,6 +247,7 @@ def association_table(
         help="The association category to get associations for, ex. biolink:GeneToPhenotypicFeatureAssociation",
     ),
     q: str = typer.Option(None, "--query", "-q"),
+    sort: List[str] = typer.Option(None, "--sort", "-s"),
     limit: int = typer.Option(5, "--limit", "-l"),
     offset: int = typer.Option(0, "--offset"),
     fmt: str = typer.Option(

backend/src/monarch_py/implementations/solr/solr_implementation.py

@@ -224,6 +224,7 @@ def get_associations(
         object_closure: str = None,
         entity: List[str] = None,
         direct: bool = None,
+        q: str = None,
         offset: int = 0,
         limit: int = 20,
     ) -> AssociationResults:
@@ -237,6 +238,7 @@ def get_associations(
             subject_closure: Filter to only associations with the specified term ID as an ancestor of the subject. Defaults to None
             object_closure: Filter to only associations with the specified term ID as an ancestor of the object. Defaults to None
             entity: Filter to only associations where the specified entities are the subject or the object. Defaults to None.
+            q: Query string to search within matches. Defaults to None.
             offset: Result offset, for pagination. Defaults to 0.
             limit: Limit results to specified number. Defaults to 20.
 
@@ -254,6 +256,7 @@ def get_associations(
             subject_closure=subject_closure,
             object_closure=object_closure,
             direct=direct,
+            q=q,
             offset=offset,
             limit=limit,
         )

backend/src/monarch_py/implementations/solr/solr_query_utils.py

@@ -56,8 +56,9 @@ def build_association_query(
     if q:
         # We don't yet have tokenization strategies for the association index, initially we'll limit searching to
         # the visible fields in an association table plus their ID equivalents and use a wildcard query for substring matching
-        query.q = f"*{q}*"
-        query.query_fields = "subject subject_label predicate object object_label"
+        query.q = q
+        query.def_type = "edismax"
+        query.query_fields = association_search_query_fields()
     if sort:
         query.sort = ", ".join(sort)
     if facet_fields:
@@ -214,3 +215,16 @@ def entity_query_fields():
     since the field list and boosts are currently the same
     """
     return "id^100 name^10 name_t^5 name_ac symbol^10 symbol_t^5 symbol_ac synonym synonym_t synonym_ac"
+
+
+def association_search_query_fields():
+    """
+    Shared field list for free text search on associations (e.g. for the association table)
+    """
+
+    return (
+        "subject subject_label^2 subject_label_t subject_closure subject_closure_label subject_closure_label_t"
+        " predicate predicate_t"
+        " object object_label^2 object_label_t object_closure object_closure_label object_closure_label_t"
+        " publications has_evidence primary_knowledge_source aggregator_knowledge_source provided_by "
+    )

backend/src/monarch_py/service/solr_service.py

@@ -26,7 +26,7 @@ def get(self, id):
     def query(self, q: SolrQuery) -> SolrQueryResult:
         url = f"{self.base_url}/{self.core.value}/select?{q.query_string()}"
         response = requests.get(url)
-
+        logger.debug(f"SolrService.query: {url}")
         data = json.loads(response.text)
         if "error" in data:
             logger.error("Solr error message: " + data["error"]["msg"])

backend/src/monarch_py/solr_cli.py

@@ -343,6 +343,7 @@ def association_table(
         help="The association category to get associations for, ex. biolink:GeneToPhenotypicFeatureAssociation",
     ),
     q: str = typer.Option(None, "--query", "-q"),
+    sort: List[str] = typer.Option(None, "--sort", "-s"),
     limit: int = typer.Option(5, "--limit", "-l"),
     offset: int = typer.Option(0, "--offset"),
     fmt: str = typer.Option(
@@ -354,7 +355,7 @@ def association_table(
     output: str = typer.Option(None, "--output", "-O", help="The path to the output file"),
 ):
     solr = get_solr(update=False)
-    response = solr.get_association_table(entity=entity, category=category, q=q, limit=limit, offset=offset)
+    response = solr.get_association_table(entity=entity, category=category, sort=sort, q=q, limit=limit, offset=offset)
     format_output(fmt, response, output)
 
 

backend/tests/fixtures/association_query_direct.py

@@ -4,7 +4,7 @@
 @pytest.fixture
 def association_query_direct():
     return {
-        "q": "*test:q*",
+        "q": "test:q",
         "rows": 100,
         "start": 100,
         "facet": True,
@@ -19,7 +19,7 @@ def association_query_direct():
             "object_closure:TEST\\:0000004",
             'subject:"TEST\\:0000005" OR object:"TEST\\:0000005"',
         ],
-        "query_fields": "subject subject_label predicate object object_label",
+        "query_fields": "subject subject_label^2 subject_label_t subject_closure subject_closure_label subject_closure_label_t predicate predicate_t object object_label^2 object_label_t object_closure object_closure_label object_closure_label_t publications has_evidence primary_knowledge_source aggregator_knowledge_source provided_by ",
         "def_type": "edismax",
         "q_op": "AND",
         "mm": "100%",

backend/tests/fixtures/association_query_indirect.py

@@ -4,7 +4,7 @@
 @pytest.fixture
 def association_query_indirect():
     return {
-        "q": "*test:q*",
+        "q": "test:q",
         "rows": 100,
         "start": 100,
         "facet": True,
@@ -19,7 +19,7 @@ def association_query_indirect():
             "object_closure:TEST\\:0000004",
             'subject:"TEST\\:0000005" OR subject_closure:"TEST\\:0000005" OR object:"TEST\\:0000005" OR object_closure:"TEST\\:0000005"',
         ],
-        "query_fields": "subject subject_label predicate object object_label",
+        "query_fields": "subject subject_label^2 subject_label_t subject_closure subject_closure_label subject_closure_label_t predicate predicate_t object object_label^2 object_label_t object_closure object_closure_label object_closure_label_t publications has_evidence primary_knowledge_source aggregator_knowledge_source provided_by ",
         "def_type": "edismax",
         "q_op": "AND",
         "mm": "100%",

backend/tests/fixtures/association_table_response.py

@@ -5,7 +5,7 @@
 def association_table_response():
     return {
         "responseHeader": {
-            "QTime": 1,
+            "QTime": 0,
             "params": {
                 "mm": "100%",
                 "q": "*:*",

backend/tests/fixtures/histopheno_response.py

@@ -5,7 +5,7 @@
 def histopheno_response():
     return {
         "responseHeader": {
-            "QTime": 4,
+            "QTime": 1,
             "params": {
                 "facet.query": [
                     'object_closure:"HP:0000924"',

backend/tests/fixtures/phenotype_explorer_compare.py

@@ -5,8 +5,8 @@
 def phenotype_explorer_compare():
     return {
         "subject_termset": {
-            "MP:0002169": {"id": "MP:0002169", "label": "no abnormal phenotype detected (MPO)"},
             "MP:0010771": {"id": "MP:0010771", "label": "integument phenotype (MPO)"},
+            "MP:0002169": {"id": "MP:0002169", "label": "no abnormal phenotype detected (MPO)"},
         },
         "object_termset": {"HP:0004325": {"id": "HP:0004325", "label": "Decreased body weight (HPO)"}},
         "subject_best_matches": {

backend/tests/integration/test_solr_association.py

@@ -82,3 +82,27 @@ def test_multi_entity_associations():
     for c in response[0].associated_categories:
         if c.counterpart_category == "biolink:Disease":
             assert c.total > 0
+
+
+@pytest.mark.parametrize("q", ["eyebrow", "thick", "Thick", "Thick eyebrow", "thick eyebrow", "Thick eyebrow (HPO)"])
+def test_association_search_partial_match(q: str):
+    si = SolrImplementation()
+    response = si.get_associations(
+        q=q, subject="MONDO:0011518", category="biolink:DiseaseToPhenotypicFeatureAssociation"
+    )
+    assert response
+    assert response.total > 0
+    assert "HP:0000574" in [item.object for item in response.items]
+
+
+@pytest.mark.parametrize("q", ["eyebrow", "thick", "Thick", "Thick eyebrow", "thick eyebrow", "Thick eyebrow (HPO)"])
+def test_association_table_search_partial_match(q: str):
+    si = SolrImplementation()
+    response = si.get_association_table(
+        entity="MONDO:0011518",
+        category="biolink:DiseaseToPhenotypicFeatureAssociation",
+        q=q,
+    )
+    assert response
+    assert response.total > 0
+    assert "HP:0000574" in [item.object for item in response.items]

frontend/fixtures/phenotype-explorer-compare.json

@@ -1,12 +1,12 @@
 {
   "subject_termset": {
-    "MP:0002169": {
-      "id": "MP:0002169",
-      "label": "no abnormal phenotype detected (MPO)"
-    },
     "MP:0010771": {
       "id": "MP:0010771",
       "label": "integument phenotype (MPO)"
+    },
+    "MP:0002169": {
+      "id": "MP:0002169",
+      "label": "no abnormal phenotype detected (MPO)"
     }
   },
   "object_termset": {