feat: add oncologic disease parameter from mondo terms

src/disease/etl/merge.py

@@ -121,7 +121,7 @@ def record_order(record: dict) -> tuple:
             merged_properties["xrefs"] = list({r["concept_id"] for r in records[1:]})
 
         set_fields = ["aliases", "associated_with"]
-        scalar_fields = ["label", "pediatric_disease"]
+        scalar_fields = ["label", "pediatric_disease", "oncologic"]
         for record in records:
             for field in set_fields:
                 if field in record:

src/disease/etl/mondo.py

@@ -186,6 +186,8 @@ def _transform_data(self) -> None:
         diseases = self._construct_dependency_set(dag, disease_root)
         peds_neoplasm_root = "MONDO:0006517"
         pediatric_diseases = self._construct_dependency_set(dag, peds_neoplasm_root)
+        cancer_root = "MONDO:0045024"
+        cancers = self._construct_dependency_set(dag, cancer_root)
 
         reader = fastobo.iter(str(self._data_file.absolute()))
         for item in tqdm(reader, ncols=80, disable=self._silent):
@@ -198,4 +200,7 @@ def _transform_data(self) -> None:
             if concept_id.upper() in pediatric_diseases:
                 params["pediatric_disease"] = True
 
+            if concept_id.upper() in cancers:
+                params["oncologic"] = True
+
             self._load_disease(params)

src/disease/etl/oncotree.py

@@ -57,6 +57,7 @@ def _add_disease(self, disease_node: dict) -> None:
             "label": disease_node["name"],
             "xrefs": [],
             "associated_with": [],
+            "oncologic": True,
         }
         refs = disease_node.get("externalReferences", [])
         for prefix, codes in refs.items():

src/disease/query.py

@@ -319,7 +319,9 @@ def _add_disease(
         :return: completed normalized response object ready to return to user
         """
         disease_obj = domain_models.Disease(
-            id=f"normalize.disease.{record['concept_id']}", label=record["label"]
+            id=f"normalize.disease.{record['concept_id']}",
+            label=record["label"],
+            extensions=[],
         )
 
         source_ids = record.get("xrefs", []) + record.get("associated_with", [])
@@ -341,12 +343,18 @@ def _add_disease(
             disease_obj.alternativeLabels = record["aliases"]
 
         if "pediatric_disease" in record and record["pediatric_disease"] is not None:
-            disease_obj.extensions = [
+            disease_obj.extensions.append(
                 entity_models.Extension(
                     name="pediatric_disease",
                     value=record["pediatric_disease"],
                 )
-            ]
+            )
+        if "oncologic" in record and record["oncologic"] is not None:
+            disease_obj.extensions.append(
+                entity_models.Extension(
+                    name="oncologic_disease", value=record["oncologic"]
+                )
+            )
 
         response["match_type"] = match_type
         response["disease"] = disease_obj

src/disease/schemas.py

@@ -107,6 +107,7 @@ class Disease(BaseModel):
     xrefs: list[StrictStr] = []
     associated_with: list[StrictStr] = []
     pediatric_disease: bool | None = None
+    oncologic: bool | None = None
 
     model_config = ConfigDict(
         json_schema_extra={

tests/unit/test_mondo.py

@@ -40,6 +40,7 @@ def neuroblastoma():
             "mesh:D009447",
         ],
         pediatric_disease=None,
+        oncologic=True,
     )
 
 
@@ -60,6 +61,7 @@ def richter_syndrome():
             "umls:C0349631",
         ],
         pediatric_disease=None,
+        oncologic=True,
     )
 
 
@@ -79,6 +81,7 @@ def pediatric_liposarcoma():
         xrefs=["DOID:5695", "ncit:C8091"],
         associated_with=["umls:C0279984"],
         pediatric_disease=True,
+        oncologic=True,
     )
 
 
@@ -94,6 +97,7 @@ def cystic_teratoma_adult():
         pediatric_disease=None,
         xrefs=["ncit:C9012", "DOID:7079"],
         associated_with=["umls:C1368888"],
+        oncologic=True,
     )
 
 
@@ -120,6 +124,7 @@ def nsclc():
             "umls:C0007131",
             "efo:0003060",
         ],
+        oncologic=True,
     )
 
 

tests/unit/test_query.py

@@ -3,7 +3,7 @@
 from datetime import datetime
 
 import pytest
-from ga4gh.core import domain_models
+from ga4gh.core import domain_models, entity_models
 
 from disease.query import InvalidParameterException, QueryHandler
 from disease.schemas import MatchType, SourceName
@@ -81,6 +81,7 @@ def neuroblastoma():
             "neural Crest tumor, malignant",
             "neuroblastoma, malignant",
         ],
+        extensions=[entity_models.Extension(name="oncologic_disease", value=True)],
     )
 
 
@@ -92,6 +93,7 @@ def skin_myo():
         id="normalize.disease.ncit:C167370",
         label="Skin Myoepithelioma",
         alternativeLabels=["Cutaneous Myoepithelioma"],
+        extensions=[entity_models.Extension(name="oncologic_disease", value=True)],
     )