Merge pull request #64 from jodyphelan/master

add tb tests
pha4ge · Oct 11, 2021 · ee85f6d · ee85f6d
2 parents a6bcd9d + 58512a2
commit ee85f6d
Show file tree

Hide file tree

Showing 7 changed files with 1,109 additions and 21 deletions.
diff --git a/hAMRonization/TBProfilerIO.py b/hAMRonization/TBProfilerIO.py
@@ -3,8 +3,7 @@
 import json
 from .Interfaces import hAMRonizedResultIterator
 
-required_metadata = ['analysis_software_version',
-                     'reference_database_version']
+required_metadata = []
 
 
 class TBProfilerIterator(hAMRonizedResultIterator):
@@ -15,15 +14,16 @@ def __init__(self, source, metadata):
 
         self.field_mapping = {
                 'filename': 'input_file_name', 
-                'gene': 'gene_symbol',
-                'gene': 'gene_name',
+                'gene_symbol': 'gene_symbol',
+                'gene_name': 'gene_name',
                 'drug': 'drug_class',
                 'type': 'genetic_variation_type',
                 'frequency': 'variant_frequency',
                 'db_name': 'reference_database_id',
                 'db_version': 'reference_database_version',
-                'tbprofiler_version': 'analysis_software_version'
-
+                'software_name': 'analysis_software_version',
+                'tbprofiler_version': 'analysis_software_version',
+                'reference_accession': 'reference_accession'
                 }
 
         super().__init__(source, self.field_mapping, self.metadata)
@@ -38,14 +38,16 @@ def parse(self, handle):
             for drug in variant["drugs"]:
                 result = {
                     'filename': handle.name,
-                    'gene': variant['gene'],
+                    'gene_symbol': variant['gene'],
                     'gene_name': variant['gene'],
                     'drug': drug['drug'],
                     'type': 'protein_variant' if variant['change'][0]=="p" else "nucleotide_variant",
                     'frequency': variant['freq'],
                     'db_name': json_obj['db_version']['name'],
                     'db_version': json_obj['db_version']['commit'],
-                    'tbprofiler_version': json_obj['tbprofiler_version']
+                    'tbprofiler_version': json_obj['tbprofiler_version'],
+                    'software_name': 'tb-profiler',
+                    'reference_accession': variant['feature_id']
                 }
                 yield self.hAMRonize(result, self.metadata)
 
diff --git a/hAMRonization/__init__.py b/hAMRonization/__init__.py
@@ -17,7 +17,7 @@
 from hAMRonization import CSStarIO
 from hAMRonization import AmrPlusPlusIO
 from hAMRonization import ResFamsIO
-
+from hAMRonization import TBProfilerIO
 
 _FormatToIterator = {
     "abricate": AbricateIO.AbricateIterator,
@@ -34,7 +34,8 @@
     "staramr": StarAmrIO.StarAmrIterator,
     "csstar": CSStarIO.CSStarIterator,
     "amrplusplus": AmrPlusPlusIO.AmrPlusPlusIterator,
-    "resfams": ResFamsIO.ResFamsIterator
+    "resfams": ResFamsIO.ResFamsIterator,
+    "tbprofiler": TBProfilerIO.TBProfilerIterator
     }
 
 _ReportFileToUse = {
@@ -52,7 +53,8 @@
     "staramr": "resfinder.tsv",
     "csstar": "OUTPUT.tsv",
     "amrplusplus": "gene.tsv",
-    "resfams": "resfams.tblout"
+    "resfams": "resfams.tblout",
+    "tbprofiler": "OUTPUT.results.json"
     }
 
 
@@ -72,6 +74,7 @@
     "amrplusplus": AmrPlusPlusIO.required_metadata,
     "resfams": ResFamsIO.required_metadata,
     "groot": GrootIO.required_metadata,
+    "tbprofiler": TBProfilerIO.required_metadata,
 }
 
 

diff --git a/hAMRonization/hAMRonizedResult.py b/hAMRonization/hAMRonizedResult.py
@@ -19,6 +19,10 @@ class hAMRonizedResult():
     reference_accession: str
     analysis_software_name: str
     analysis_software_version: str
+
+    # variant specific optional fields
+    variant_frequency: float = None
+    genetic_variation_type: str = None
 
     # optional fields
     sequence_identity: float = None