wip

pyproject.toml

@@ -26,10 +26,10 @@ dependencies = [
     "ga4gh.vrs==2.0.0a13",
     #"ga4gh.cat_vrs~=0.1.0",
     #"ga4gh.va_spec~=0.1.0",
-    "gene-normalizer[etl]~=0.6.0",
+    "gene-normalizer[etl]==0.6.0",
     "variation-normalizer~=0.12.0",
-    "disease-normalizer[etl]~=0.7.0",
-    "thera-py[etl]~=0.7.1",
+    "disease-normalizer[etl]==0.7.0",
+    "thera-py[etl]==0.7.1",
     "civicpy~=3.1",
     "requests",
     "pydantic==2.*",

src/metakb/transformers/civic.py

@@ -248,12 +248,12 @@ async def transform(self, harvested_data: CivicHarvestedData) -> None:
 
         # Get all variant IDs from supported molecular profiles
         vids = {
-            mp_id_to_v_id_mapping[e["molecular_profile_id"]]
+            mp_id_to_v_id_mapping.get(e["molecular_profile_id"])
             for e in evidence_items
             if e["molecular_profile_id"]
         }
         vids |= {
-            mp_id_to_v_id_mapping[assertion["molecular_profile_id"]]
+            mp_id_to_v_id_mapping.get(assertion["molecular_profile_id"])
             for assertion in assertions
             if assertion["molecular_profile_id"]
         }
@@ -336,6 +336,8 @@ def _add_variant_study_stmt(
 
             if record["amp_level"]:
                 classification = self._get_classification(record["amp_level"])
+                if not classification:
+                    return
 
             evidence_lines = []
             evidence_ids = []
@@ -347,14 +349,10 @@ def _add_variant_study_stmt(
                     evidence_lines.append(
                         EvidenceLine(
                             hasEvidenceItems=[evidence_item],
-                            directionOfEvidenceProvided=Direction.SUPPORTS,  # TODO: Is this always supports?
+                            directionOfEvidenceProvided=Direction.SUPPORTS,
                         )
                     )
 
-            # TODO: Figure out how to handle cases where CIViC evidence items can't be processed. Is this what we want?
-            if evidence_ids:
-                extensions.append(Extension(name="evidence_ids", value=evidence_ids))
-
         record_type = record[f"{record_prefix}_type"]
 
         # Get predicate
@@ -472,20 +470,10 @@ def _get_classification(amp_level: str) -> MappableConcept | None:
         else:
             pattern = re.compile(r"TIER_(?P<tier>[IV]+)(?:_LEVEL_(?P<level>[A-D]))?")
             match = pattern.match(amp_level).groupdict()
-            primary_code = f"{match['tier']}{match['level'] or ''}"
+            primary_code = f"Tier {match['tier']}"
             classification = MappableConcept(
-                conceptType="Guideline",
                 primaryCode=primary_code,
-                mappings=[
-                    ConceptMapping(
-                        relation=Relation.EXACT_MATCH,
-                        coding=Coding(
-                            system="AMP/ASCO/CAP",
-                            systemVersion="2017",
-                            code=code(primary_code),
-                        ),
-                    )
-                ],
+                extensions=[Extension(name="civic_amp_level", value=amp_level)],
             )
         return classification
 

tests/conftest.py

@@ -1104,19 +1104,6 @@ def civic_aid6_statement(
         "id": "civic.aid:6",
         "description": "L858R is among the most common sensitizing EGFR mutations in NSCLC, and is assessed via DNA mutational analysis, including Sanger sequencing and next generation sequencing methods. Tyrosine kinase inhibitor afatinib is FDA approved as a first line systemic therapy in NSCLC with sensitizing EGFR mutation (civic.EID:2997).",
         "type": "Statement",
-        "extensions": [
-            {
-                "name": "evidence_ids",
-                "value": [
-                    "civic.eid:2997",
-                    "civic.eid:2629",
-                    "civic.eid:982",
-                    "civic.eid:968",
-                    "civic.eid:883",
-                    "civic.eid:879",
-                ],
-            }
-        ],
         "specifiedBy": civic_method,
         "proposition": {
             "type": "VariantTherapeuticResponseProposition",
@@ -1129,18 +1116,8 @@ def civic_aid6_statement(
         },
         "direction": "supports",
         "classification": {
-            "conceptType": "Guideline",
-            "primaryCode": "IA",
-            "mappings": [
-                {
-                    "coding": {
-                        "system": "AMP/ASCO/CAP",
-                        "systemVersion": "2017",
-                        "code": "IA",
-                    },
-                    "relation": "exactMatch",
-                }
-            ],
+            "primaryCode": "Tier I",
+            "extensions": [{"name": "civic_amp_level", "value": "TIER_I_LEVEL_A"}],
         },
         "hasEvidenceLines": [
             {