fix docstring language + enforce unique ontology list in ontology_info

api/python/src/cellxgene_ontology_guide/supported_versions.py

@@ -58,8 +58,9 @@ class CXGSchema:
     supported_ontologies: Dict[str, Any]
     """A dictionary of supported ontologies for the schema version."""
     imported_ontologies: Dict[str, str]
-    """A dictionary of ontologies with terms imported into supported ontologies, mapped to the supported ontology
-    importing them."""
+    """In our supported ontologies, the CxG schema can support terms imported from different ontologies. 
+    This dictionary maps these 'additional ontologies' to their supported ontology name. For example, 
+    for ZFS ontology terms imported into the ZFA ontology, imported_ontologies would be {"ZFS":"ZFA", ...}"""
     ontology_file_names: Dict[str, str]
     """A dictionary of ontology names and their corresponding file names."""
 

asset-schemas/ontology_info_schema.json

@@ -14,35 +14,12 @@
         },
         "ontologies": {
           "type": "object",
-          "properties": {
-            "CL": {
-              "$ref": "#/definitions/ontologyEntry"
-            },
-            "EFO": {
-              "$ref": "#/definitions/ontologyEntry"
-            },
-            "HANCESTRO": {
-              "$ref": "#/definitions/ontologyEntry"
-            },
-            "HsapDv": {
-              "$ref": "#/definitions/ontologyEntry"
-            },
-            "MONDO": {
-              "$ref": "#/definitions/ontologyEntry"
-            },
-            "MmusDv": {
-              "$ref": "#/definitions/ontologyEntry"
-            },
-            "NCBITaxon": {
-              "$ref": "#/definitions/ontologyEntry"
-            },
-            "UBERON": {
-              "$ref": "#/definitions/ontologyEntry"
-            },
-            "PATO": {
+          "patternProperties": {
+            "^[A-Za-z0-9]+$": {
               "$ref": "#/definitions/ontologyEntry"
             }
-          }
+          },
+          "additionalProperties": false
         }
       },
       "required": [

tools/ontology-builder/src/validate_json_schemas.py

@@ -67,12 +67,31 @@ def verify_json(schema_file_name: str, json_file_name: str, registry: Registry)
 
     try:
         validate(instance=data, schema=schema, registry=registry)
+        # custom logic for ontology_info definition
+        if "ontology_info" in schema_file_name:
+            validate_unique_ontologies(data)
     except Exception:
         logger.exception(f"Error validating {json_file_name} against {schema_file_name}")
         return False
     return True
 
 
+def validate_unique_ontologies(data):
+    """"
+    Custom validation logic to check that all ontologies (including additional_ontologies) defined in ontology_info
+    are unique across entries
+    """
+    for schema_version, version_info in data.items():
+        all_ontologies = []
+        for ontology, ontology_info in version_info["ontologies"].items():
+            all_ontologies.append(ontology)
+            all_ontologies.extend(ontology_info.get("additional_ontologies", []))
+        if len(all_ontologies) != len(set(all_ontologies)):
+            logger.error("Ontology entries must be unique across all ontology entries, including "
+                         f"additional_ontologies. Duplicates found in definition for {schema_version}")
+            raise ValueError
+
+
 def main(path: str = env.ONTOLOGY_ASSETS_DIR) -> None:
     """
     Verify the curated JSON lists match their respective JSON schema in asset-schemas

tools/ontology-builder/tests/test_validate_json_schemas.py

@@ -1,5 +1,6 @@
 import gzip
 import json
+import os
 
 import pytest
 from referencing import Resource
@@ -108,3 +109,84 @@ def test_invalid_schema(self, schema_file_fixture, tmpdir, registry_fixture):
 
         # Assert validation fails due to invalid schema
         assert verify_json(schema_file_fixture, str(json_file), registry_fixture) is False
+
+
+class TestVerifyJsonCustomLogic:
+    @pytest.fixture
+    def ontology_info_schema_file_fixture(self, tmpdir):
+        ontology_info_schema = os.path.join(
+            os.path.realpath(__file__).rsplit("/", maxsplit=4)[0], "asset-schemas", "ontology_info_schema.json"
+        )
+        with open(ontology_info_schema, "r") as f:
+            schema_data = json.load(f)
+        schema_file = tmpdir.join("ontology_info_schema.json")
+        with open(str(schema_file), "w") as f:
+            json.dump(schema_data, f)
+        return str(schema_file)
+
+    @pytest.fixture
+    def ontology_info_registry_fixture(self, ontology_info_schema_file_fixture, tmpdir):
+        return register_schemas(tmpdir)
+
+    @pytest.fixture
+    def ontology_info_json_data(self):
+        return {
+            "2.0.0": {
+                "ontologies": {
+                    "A": {
+                        "version": "v1",
+                        "source": "https://example.org/ontology/download",
+                        "filename": "a.owl",
+                        "additional_ontologies": ["C"],
+                    },
+                    "B": {
+                        "version": "v2",
+                        "source": "https://example.org/ontology/download",
+                        "filename": "b.owl",
+                        "additional_ontologies": ["D"],
+                    },
+                }
+            },
+            "1.0.0": {
+                "ontologies": {
+                    "A": {
+                        "version": "v1",
+                        "source": "https://example.org/ontology/download",
+                        "filename": "a.owl",
+                    },
+                    "B": {
+                        "version": "v1",
+                        "source": "https://example.org/ontology/download",
+                        "filename": "b.owl",
+                    },
+                }
+            },
+        }
+
+    def test_validate_unique_ontologies(
+        self, ontology_info_json_data, ontology_info_schema_file_fixture, tmpdir, ontology_info_registry_fixture
+    ):
+        json_file = tmpdir.join("ontology_info.json")
+        with open(str(json_file), "w") as f:
+            json.dump(ontology_info_json_data, f)
+
+        # Assert validation passes
+        assert verify_json(ontology_info_schema_file_fixture, str(json_file), ontology_info_registry_fixture) is True
+
+    @pytest.mark.parametrize("additional_ontologies", [["C"], ["A"]])
+    def test_validate_unique_ontologies__invalid(
+        self,
+        additional_ontologies,
+        ontology_info_json_data,
+        ontology_info_schema_file_fixture,
+        tmpdir,
+        ontology_info_registry_fixture,
+    ):
+        # Create invalid JSON data
+        ontology_info_json_data["2.0.0"]["ontologies"]["B"]["additional_ontologies"] = additional_ontologies
+        json_file = tmpdir.join("ontology_info.json")
+        with open(str(json_file), "w") as f:
+            json.dump(ontology_info_json_data, f)
+
+        # Assert validation fails
+        assert verify_json(ontology_info_schema_file_fixture, str(json_file), ontology_info_registry_fixture) is False