feat!: add support for transforming civic diagnostic evidence items (#…

…414)

close #241

* MetaKB now supports `VariantDiagnosticStudyStatement`s in addition to
`VariantTherapeuticResponseStudyStatement` and
`VariantPrognosticStudyStatement`

src/metakb/query.py

@@ -15,6 +15,7 @@
 )
 from ga4gh.core.entity_models import Coding, Document, Extension, Method
 from ga4gh.va_spec.profiles.var_study_stmt import (
+    VariantDiagnosticStudyStatement,
     VariantPrognosticStudyStatement,
     VariantTherapeuticResponseStudyStatement,
 )
@@ -69,6 +70,14 @@ class TherapeuticProcedureType(str, Enum):
     SUBSTITUTES = "TherapeuticSubstituteGroup"
 
 
+# Statement types to corresponding class mapping
+STMT_TYPE_TO_CLASS = {
+    "VariantDiagnosticStudyStatement": VariantDiagnosticStudyStatement,
+    "VariantPrognosticStudyStatement": VariantPrognosticStudyStatement,
+    "VariantTherapeuticResponseStudyStatement": VariantTherapeuticResponseStudyStatement,
+}
+
+
 def _deserialize_field(node: dict, field_name: str) -> None | dict:
     """Deserialize JSON blob property.
 
@@ -493,8 +502,8 @@ def _get_nested_stmts(self, statement_nodes: list[Node]) -> list[dict]:
 
     def _get_nested_stmt(self, stmt_node: Node) -> dict:
         """Get information related to a statement
-        Only VariantTherapeuticResponseStudyStatement and VariantPrognosticStudyStatement
-        are supported at the moment
+        Only VariantTherapeuticResponseStudyStatement, VariantPrognosticStudyStatement,
+        and VariantDiagnosticStudyStatement are supported at the moment
 
         :param stmt_node: Neo4j graph node for statement
         :return: Nested statement
@@ -503,15 +512,14 @@ def _get_nested_stmt(self, stmt_node: Node) -> dict:
         if study_stmt_type not in {
             "VariantTherapeuticResponseStudyStatement",
             "VariantPrognosticStudyStatement",
+            "VariantDiagnosticStudyStatement",
         }:
             return {}
 
-        if study_stmt_type == "VariantPrognosticStudyStatement":
-            study_stmt_cls = VariantPrognosticStudyStatement
-            condition_key = "objectCondition"
-        else:
-            study_stmt_cls = VariantTherapeuticResponseStudyStatement
+        if study_stmt_type == "VariantTherapeuticResponseStudyStatement":
             condition_key = "conditionQualifier"
+        else:
+            condition_key = "objectCondition"
 
         params = {
             condition_key: None,
@@ -559,7 +567,7 @@ def _get_nested_stmt(self, stmt_node: Node) -> dict:
             else:
                 logger.warning("relation type not supported: %s", rel_type)
 
-        return study_stmt_cls(**params).model_dump()
+        return STMT_TYPE_TO_CLASS[study_stmt_type](**params).model_dump()
 
     @staticmethod
     def _get_vicc_normalizer_extension(node: dict) -> ViccNormalizerDataExtension:
@@ -917,10 +925,6 @@ async def batch_search_statements(
             statement_nodes = [r[0] for r in result]
         response.statement_ids = [n["id"] for n in statement_nodes]
         stmts = self._get_nested_stmts(statement_nodes)
-        response.statements = [
-            VariantTherapeuticResponseStudyStatement(**s)
-            if s["type"] == "VariantTherapeuticResponseStudyStatement"
-            else VariantPrognosticStudyStatement(**s)
-            for s in stmts
-        ]
+
+        response.statements = [STMT_TYPE_TO_CLASS[s["type"]](**s) for s in stmts]
         return response

src/metakb/schemas/api.py

@@ -3,6 +3,7 @@
 from typing import Literal
 
 from ga4gh.va_spec.profiles.var_study_stmt import (
+    VariantDiagnosticStudyStatement,
     VariantPrognosticStudyStatement,
     VariantTherapeuticResponseStudyStatement,
 )
@@ -48,7 +49,9 @@ class SearchStatementsService(BaseModel):
     warnings: list[StrictStr] = []
     statement_ids: list[StrictStr] = []
     statements: list[
-        VariantTherapeuticResponseStudyStatement | VariantPrognosticStudyStatement
+        VariantTherapeuticResponseStudyStatement
+        | VariantPrognosticStudyStatement
+        | VariantDiagnosticStudyStatement
     ] = []
     service_meta_: ServiceMeta
 
@@ -73,6 +76,8 @@ class BatchSearchStatementsService(BaseModel):
     warnings: list[StrictStr] = []
     statement_ids: list[StrictStr] = []
     statements: list[
-        VariantTherapeuticResponseStudyStatement | VariantPrognosticStudyStatement
+        VariantTherapeuticResponseStudyStatement
+        | VariantPrognosticStudyStatement
+        | VariantDiagnosticStudyStatement
     ] = []
     service_meta_: ServiceMeta

src/metakb/transformers/base.py

@@ -25,6 +25,7 @@
 )
 from ga4gh.core.entity_models import Coding, Document, Extension, Method
 from ga4gh.va_spec.profiles.var_study_stmt import (
+    VariantDiagnosticStudyStatement,
     VariantPrognosticStudyStatement,
     VariantTherapeuticResponseStudyStatement,
 )
@@ -111,7 +112,9 @@ class TransformedData(BaseModel):
     """Define model for transformed data"""
 
     statements: list[
-        VariantTherapeuticResponseStudyStatement | VariantPrognosticStudyStatement
+        VariantTherapeuticResponseStudyStatement
+        | VariantPrognosticStudyStatement
+        | VariantDiagnosticStudyStatement
     ] = []
     categorical_variants: list[CategoricalVariant] = []
     variations: list[Allele] = []

src/metakb/transformers/civic.py

@@ -22,8 +22,10 @@
 )
 from ga4gh.va_spec.profiles.var_study_stmt import (
     AlleleOriginQualifier,
+    DiagnosticPredicate,
     PrognosticPredicate,
     TherapeuticResponsePredicate,
+    VariantDiagnosticStudyStatement,
     VariantPrognosticStudyStatement,
     VariantTherapeuticResponseStudyStatement,
 )
@@ -91,6 +93,8 @@
     "RESISTANCE": TherapeuticResponsePredicate.RESISTANCE,
     "POOR_OUTCOME": PrognosticPredicate.WORSE_OUTCOME,
     "BETTER_OUTCOME": PrognosticPredicate.BETTER_OUTCOME,
+    "POSITIVE": DiagnosticPredicate.INCLUSIVE,
+    "NEGATIVE": DiagnosticPredicate.EXCLUSIVE,
 }
 
 
@@ -121,6 +125,7 @@ class _CivicEvidenceType(str, Enum):
 
     PREDICTIVE = "PREDICTIVE"
     PROGNOSTIC = "PROGNOSTIC"
+    DIAGNOSTIC = "DIAGNOSTIC"
 
 
 class _VariationCache(BaseModel):
@@ -359,8 +364,10 @@ def _add_variant_study_stmt(
         if evidence_type == _CivicEvidenceType.PREDICTIVE:
             params["objectTherapeutic"] = civic_therapeutic
             statement = VariantTherapeuticResponseStudyStatement(**params)
-        else:
+        elif evidence_type == _CivicEvidenceType.PROGNOSTIC:
             statement = VariantPrognosticStudyStatement(**params)
+        else:
+            statement = VariantDiagnosticStudyStatement(**params)
 
         self.processed_data.statements.append(statement)