feat: add oncologic disease parameter from mondo terms (#200)

src/disease/database/postgresql.py

@@ -318,6 +318,7 @@ def _format_source_record(self, source_row: tuple) -> dict:
             "src_name": source_row[5],
             "merge_ref": source_row[6],
             "pediatric_disease": source_row[7],
+            "oncologic_disease": source_row[8],
             "item_type": RecordType.IDENTITY.value,
         }
         return {k: v for k, v in disease_record.items() if v}
@@ -354,6 +355,7 @@ def _format_merged_record(self, merged_row: tuple) -> dict:
             "associated_with": merged_row[3],
             "xrefs": merged_row[4],
             "pediatric_disease": merged_row[5],
+            "oncologic_disease": merged_row[6],
             "item_type": RecordType.MERGER.value,
         }
         return {k: v for k, v in merged_record.items() if v}
@@ -532,8 +534,8 @@ def add_source_metadata(self, src_name: SourceName, meta: SourceMeta) -> None:
         self.conn.commit()
 
     _add_record_query = b"""
-        INSERT INTO disease_concepts (concept_id, source, pediatric_disease)
-        VALUES (%s, %s, %s);
+        INSERT INTO disease_concepts (concept_id, source, pediatric_disease, oncologic_disease)
+        VALUES (%s, %s, %s, %s);
     """
     _insert_label_query = (
         b"INSERT INTO disease_labels (label, concept_id) VALUES (%s, %s)"
@@ -555,7 +557,12 @@ def add_record(self, record: dict, src_name: SourceName) -> None:
             try:
                 cur.execute(
                     self._add_record_query,
-                    [concept_id, src_name.value, record.get("pediatric_disease")],
+                    [
+                        concept_id,
+                        src_name.value,
+                        record.get("pediatric_disease"),
+                        record.get("oncologic_disease"),
+                    ],
                 )
                 cur.execute(self._insert_label_query, [record["label"], concept_id])
                 for a in record.get("aliases", []):
@@ -571,9 +578,9 @@ def add_record(self, record: dict, src_name: SourceName) -> None:
 
     _add_merged_record_query = b"""
     INSERT INTO disease_merged (
-        concept_id, label, aliases, associated_with, xrefs, pediatric_disease
+        concept_id, label, aliases, associated_with, xrefs, pediatric_disease, oncologic_disease
     )
-    VALUES (%s, %s, %s, %s, %s, %s);
+    VALUES (%s, %s, %s, %s, %s, %s, %s);
     """
 
     def add_merged_record(self, record: dict) -> None:
@@ -591,6 +598,7 @@ def add_merged_record(self, record: dict) -> None:
                     record.get("associated_with"),
                     record.get("xrefs"),
                     record.get("pediatric_disease"),
+                    record.get("oncologic_disease"),
                 ],
             )
             self.conn.commit()

src/disease/database/postgresql/create_record_lookup_view.sql

@@ -7,6 +7,7 @@ SELECT dc.concept_id,
        dc.source,
        dc.merge_ref,
        dc.pediatric_disease,
+       dc.oncologic_disease,
        lower(dc.concept_id) AS concept_id_lowercase
 FROM disease_concepts dc
 FULL JOIN (

src/disease/database/postgresql/create_tables.sql

@@ -16,12 +16,14 @@ CREATE TABLE disease_merged (
     aliases TEXT [],
     associated_with TEXT [],
     xrefs TEXT [],
-    pediatric_disease BOOLEAN
+    pediatric_disease BOOLEAN,
+    oncologic_disease BOOLEAN
 );
 CREATE TABLE disease_concepts (
     concept_id VARCHAR(127) PRIMARY KEY,
     source VARCHAR(127) NOT NULL REFERENCES disease_sources (name),
     pediatric_disease BOOLEAN,
+    oncologic_disease BOOLEAN,
     merge_ref VARCHAR(127) REFERENCES disease_merged (concept_id)
 );
 CREATE TABLE disease_labels (

src/disease/database/postgresql/delete_normalized_concepts.sql

@@ -10,7 +10,8 @@ CREATE TABLE disease_merged (
     aliases TEXT [],
     associated_with TEXT [],
     xrefs TEXT [],
-    pediatric_disease BOOLEAN
+    pediatric_disease BOOLEAN,
+    oncologic_disease BOOLEAN
 );
 ALTER TABLE disease_concepts ADD CONSTRAINT disease_concepts_merge_ref_fkey
     FOREIGN KEY (merge_ref) REFERENCES disease_merged (concept_id);

src/disease/etl/base.py

@@ -114,8 +114,9 @@ def _load_disease(self, disease: dict) -> None:
                 else:
                     del disease[attr_type]
 
-        if "pediatric_disease" in disease and disease["pediatric_disease"] is None:
-            del disease["pediatric_disease"]
+        for field in ("pediatric_disease", "oncologic_disease"):
+            if field in disease and disease[field] is None:
+                del disease[field]
 
         self._database.add_record(disease, self._src_name)
         if self._store_ids:

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_disease"]
         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_disease"] = 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_disease": 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", [])
@@ -340,13 +342,12 @@ def _add_disease(
         if "aliases" in record:
             disease_obj.alternativeLabels = record["aliases"]
 
-        if "pediatric_disease" in record and record["pediatric_disease"] is not None:
-            disease_obj.extensions = [
-                entity_models.Extension(
-                    name="pediatric_disease",
-                    value=record["pediatric_disease"],
+        for field in ("pediatric_disease", "oncologic_disease"):
+            value = record.get(field)
+            if value is not None:
+                disease_obj.extensions.append(
+                    entity_models.Extension(name=field, value=value)
                 )
-            ]
 
         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_disease: bool | None = None
 
     model_config = ConfigDict(
         json_schema_extra={
@@ -123,6 +124,7 @@ class Disease(BaseModel):
                 "xrefs": [],
                 "associated_with": ["umls:C0019562"],
                 "pediatric_disease": None,
+                "oncologic_disease": None,
             }
         }
     )

tests/conftest.py

@@ -144,6 +144,10 @@ def _compare_records(actual: Disease, fixt: Disease):
     if (actual.pediatric_disease is not None) and (fixt.pediatric_disease is not None):
         assert actual.pediatric_disease == fixt.pediatric_disease
 
+    assert (actual.oncologic_disease is None) == (fixt.oncologic_disease is None)
+    if (actual.oncologic_disease is not None) and (fixt.oncologic_disease is not None):
+        assert actual.oncologic_disease == fixt.oncologic_disease
+
 
 @pytest.fixture(scope="session")
 def compare_records():

tests/unit/test_merge.py

@@ -68,7 +68,7 @@ def neuroblastoma():
             "umls:C2751421",
             "umls:CN205405",
         ],
-        "pediatric": None,
+        "oncologic_disease": True,
     }
 
 
@@ -107,6 +107,7 @@ def lnscc():
             "kegg.disease:05223",
         ],
         "item_type": "merger",
+        "oncologic_disease": True,
     }
 
 
@@ -131,6 +132,7 @@ def richter():
             "icd10.cm:C91.1",
         ],
         "item_type": "merger",
+        "oncologic_disease": True,
     }
 
 
@@ -150,6 +152,7 @@ def ped_liposarcoma():
         ],
         "associated_with": ["umls:C0279984"],
         "pediatric_disease": True,
+        "oncologic_disease": True,
         "item_type": "merger",
     }
 
@@ -168,6 +171,7 @@ def teratoma():
         ],
         "associated_with": ["icdo:9080/0", "umls:C1368888"],
         "item_type": "merger",
+        "oncologic_disease": True,
     }
 
 
@@ -238,6 +242,10 @@ def compare_merged_records(actual, fixture):
     if "pediatric_disease" in actual or "pediatric_disease" in fixture:
         assert actual["pediatric_disease"] == fixture["pediatric_disease"]
 
+    assert ("oncologic_disease" in actual) == ("oncologic_disease" in fixture)
+    if "oncologic_disease" in actual or "oncologic_disease" in fixture:
+        assert actual["oncologic_disease"] == fixture["oncologic_disease"]
+
 
 def test_generate_merged_record(
     merge_instance,

tests/unit/test_mondo.py

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

tests/unit/test_omim.py

@@ -32,6 +32,7 @@ def mafd2():
         xrefs=[],
         associated_with=[],
         pediatric_disease=None,
+        oncologic_disease=None,
     )
 
 
@@ -53,6 +54,7 @@ def acute_ll():
         xrefs=[],
         associated_with=[],
         pediatric_disease=None,
+        oncologic_disease=None,
     )
 
 
@@ -66,6 +68,7 @@ def lall():
         xrefs=[],
         associated_with=[],
         pediatric_disease=None,
+        oncologic_disease=None,
     )
 
 

tests/unit/test_oncotree.py

@@ -22,6 +22,7 @@ def neuroblastoma():
         xrefs=["ncit:C3270"],
         associated_with=["umls:C0027819"],
         pediatric_disease=None,
+        oncologic_disease=True,
     )
 
 
@@ -35,6 +36,7 @@ def nsclc():
         xrefs=["ncit:C2926"],
         associated_with=["umls:C0007131"],
         pediatric_disease=None,
+        oncologic_disease=True,
     )
 
 
@@ -48,6 +50,7 @@ def ipn():
         xrefs=[],
         associated_with=[],
         pediatric_disease=None,
+        oncologic_disease=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=[],
     )
 
 
@@ -130,6 +132,7 @@ def mafd2():
                 "relation": "relatedMatch",
             },
         ],
+        extensions=[],
     )
 
 
@@ -179,9 +182,16 @@ def get_extension(extensions, name):
         ped_fixture = get_extension(ext_fixture, "pediatric_disease")
         assert (ped_actual is None) == (ped_fixture is None)
         if ped_actual and ped_fixture:
-            assert set(ped_actual.value) == set(ped_fixture.value)
+            assert ped_actual.value == ped_fixture.value
             assert ped_actual.value
 
+        onco_actual = get_extension(ext_actual, "oncologic_disease")
+        onco_fixture = get_extension(ext_fixture, "oncologic_disease")
+        assert (onco_actual is None) == (onco_fixture is None)
+        if onco_actual and onco_fixture:
+            assert onco_actual.value == onco_fixture.value
+            assert onco_actual.value
+
 
 def test_query(query_handler):
     """Test that query returns properly-structured response."""