feat: refer to ontology source filenames in ontology_info and return …

…that in get_ontology_download_url (#106)
chanzuckerberg · Mar 13, 2024 · ff9d826 · ff9d826
1 parent 341d418
commit ff9d826
Show file tree

Hide file tree

Showing 17 changed files with 27 additions and 66 deletions.
diff --git a/api/python/pyproject.toml b/api/python/pyproject.toml
@@ -11,7 +11,7 @@ authors = [
 ]
 license = { file = "LICENSE" }
 readme = "README.md"
-requires-python = "~= 3.11"
+requires-python = "~= 3.10"
 dependencies = ["semantic_version==2.10.0"]
 
 [project.optional-dependencies]

diff --git a/api/python/src/cellxgene_ontology_guide/entities.py b/api/python/src/cellxgene_ontology_guide/entities.py
@@ -15,28 +15,3 @@ class Ontology(Enum):
     MmusDv = "mmusdv"
     PATO = "pato"
     NCBITaxon = "ncbitaxon"
-
-
-class OntologyVariant(Enum):
-    """
-    Enum for the standard set of ontology variants. Each is curated for a specific purpose.
-
-    See https://oboacademy.github.io/obook/explanation/owl-format-variants/ for more information on the distinction
-    and use-cases for each variant.
-    """
-
-    FULL = "full"
-    BASE = "base"
-    SIMPLE = "simple"
-    BASIC = "basic"
-
-
-class OntologyFileType(Enum):
-    """
-    Enum for the standard set of ontology file types. Each requires different parsing tools, but relay the same
-    information.
-    """
-
-    OWL = "owl"
-    OBO = "obo"
-    JSON = "json"
diff --git a/api/python/src/cellxgene_ontology_guide/ontology_parser.py b/api/python/src/cellxgene_ontology_guide/ontology_parser.py
@@ -1,7 +1,7 @@
 import re
 from typing import Any, Dict, Iterable, List, Union
 
-from cellxgene_ontology_guide.entities import Ontology, OntologyFileType, OntologyVariant
+from cellxgene_ontology_guide.entities import Ontology
 from cellxgene_ontology_guide.supported_versions import CXGSchema
 
 
@@ -274,26 +274,17 @@ def get_term_label(self, term_id: str) -> str:
         label: str = self.cxg_schema.ontology(ontology_name)[term_id]["label"]
         return label
 
-    def get_ontology_download_url(
-        self, ontology: Ontology, ontology_filetype: OntologyFileType, ontology_variant: OntologyVariant = None
-    ) -> str:
+    def get_ontology_download_url(self, ontology: Ontology) -> str:
         """
-        Get the download URL for a given ontology file. If the ontology_variant is not provided, the default ontology
-        file will be returned.
+        Get the download URL for a given ontology file.
 
         Examples:
-        get_ontology_download_url("CL", "owl") -> "http://example.com/2024-01-01/cl.owl"
-        get_ontology_download_url("CL", "obo", "base") -> "http://example.com/2024-01-01/cl-base.obo"
+        get_ontology_download_url("CL") -> "http://example.com/2024-01-01/cl.owl"
 
         :param ontology: Ontology enum of the ontology to fetch
-        :param ontology_filetype: OntologyFileType enum of the ontology file type to fetch
-        :param ontology_variant: OntologyVariant enum of the ontology variant to fetch
         :return: str download URL for the requested ontology file
         """
         source_url = self.cxg_schema.supported_ontologies[ontology.name]["source"]
         version = self.cxg_schema.supported_ontologies[ontology.name]["version"]
-        return (
-            f"{source_url}/{version}/{ontology.value}-{ontology_variant.value}.{ontology_filetype.value}"
-            if ontology_variant
-            else f"{source_url}/{version}/{ontology.value}.{ontology_filetype.value}"
-        )
+        filename = self.cxg_schema.supported_ontologies[ontology.name]["filename"]
+        return f"{source_url}/{version}/{filename}"
diff --git a/api/python/tests/test_ontology_parser.py b/api/python/tests/test_ontology_parser.py
@@ -1,7 +1,7 @@
 from unittest.mock import patch
 
 import pytest
-from cellxgene_ontology_guide.entities import Ontology, OntologyFileType, OntologyVariant
+from cellxgene_ontology_guide.entities import Ontology
 from cellxgene_ontology_guide.ontology_parser import OntologyParser
 from cellxgene_ontology_guide.supported_versions import CXGSchema
 
@@ -44,7 +44,7 @@ def ontology_dict():
 @pytest.fixture
 def mock_CXGSchema(ontology_dict, mock_load_supported_versions, mock_load_ontology_file):
     mock_load_supported_versions.return_value = {
-        "v5.0.0": {"CL": {"version": "2024-01-01", "source": "http://example.com", "filetype": "owl"}}
+        "v5.0.0": {"CL": {"version": "2024-01-01", "source": "http://example.com", "filename": "cl.owl"}}
     }
     cxg_schema = CXGSchema()
     cxg_schema.ontology_file_names = {"CL": "CL-ontology-2024-01-01.json.gz"}
@@ -212,11 +212,4 @@ def test_get_distance_between_terms(ontology_parser):
 
 
 def test_get_ontology_download_url(ontology_parser):
-    assert (
-        ontology_parser.get_ontology_download_url(Ontology.CL, OntologyFileType.OWL)
-        == "http://example.com/2024-01-01/cl.owl"
-    )
-    assert (
-        ontology_parser.get_ontology_download_url(Ontology.CL, OntologyFileType.OBO, OntologyVariant.BASE)
-        == "http://example.com/2024-01-01/cl-base.obo"
-    )
+    assert ontology_parser.get_ontology_download_url(Ontology.CL) == "http://example.com/2024-01-01/cl.owl"
diff --git a/artifact-schemas/ontology_info_schema.json b/artifact-schemas/ontology_info_schema.json
@@ -52,15 +52,15 @@
           "format": "uri",
           "description": "Source to find and download canonical ontology version for this data release"
         },
-        "filetype": {
+        "filename": {
           "type": "string",
-          "description": "filetype used to build generated artifacts for this ontology data release"
+          "description": "name of ontology file used to build generated artifacts for this ontology data release"
         }
       },
       "required": [
         "version",
         "source",
-        "filetype"
+        "filename"
       ],
       "additionalProperties": false
     }

diff --git a/ontology-assets/CL-ontology-v2024-01-04.json.gz b/ontology-assets/CL-ontology-v2024-01-04.json.gz
diff --git a/ontology-assets/EFO-ontology-v3.62.0.json.gz b/ontology-assets/EFO-ontology-v3.62.0.json.gz
diff --git a/ontology-assets/HANCESTRO-ontology-3.0.json.gz b/ontology-assets/HANCESTRO-ontology-3.0.json.gz
diff --git a/ontology-assets/HsapDv-ontology-11.json.gz b/ontology-assets/HsapDv-ontology-11.json.gz
diff --git a/ontology-assets/MONDO-ontology-v2024-01-03.json.gz b/ontology-assets/MONDO-ontology-v2024-01-03.json.gz
diff --git a/ontology-assets/MmusDv-ontology-9.json.gz b/ontology-assets/MmusDv-ontology-9.json.gz
diff --git a/ontology-assets/NCBITaxon-ontology-v2023-06-20.json.gz b/ontology-assets/NCBITaxon-ontology-v2023-06-20.json.gz
diff --git a/ontology-assets/PATO-ontology-v2023-05-18.json.gz b/ontology-assets/PATO-ontology-v2023-05-18.json.gz
diff --git a/ontology-assets/UBERON-ontology-v2024-01-18.json.gz b/ontology-assets/UBERON-ontology-v2024-01-18.json.gz
diff --git a/ontology-assets/ontology_info.json b/ontology-assets/ontology_info.json
@@ -3,47 +3,47 @@
     "CL": {
       "version": "v2024-01-04",
       "source": "https://github.com/obophenotype/cell-ontology/releases/download",
-      "filetype": "owl"
+      "filename": "cl.owl"
     },
     "EFO": {
       "version": "v3.62.0",
       "source": "https://github.com/EBISPOT/efo/releases/download",
-      "filetype": "owl"
+      "filename": "efo.owl"
     },
     "HANCESTRO": {
       "version": "3.0",
       "source": "https://github.com/EBISPOT/hancestro/raw",
-      "filetype": "owl"
+      "filename": "hancestro.owl"
     },
     "HsapDv": {
       "version": "11",
       "source": "http://aber-owl.net/media/ontologies/HSAPDV",
-      "filetype": "owl"
+      "filename": "hsapdv.owl"
     },
     "MONDO": {
       "version": "v2024-01-03",
       "source": "https://github.com/monarch-initiative/mondo/releases/download",
-      "filetype": "owl"
+      "filename": "mondo.owl"
     },
     "MmusDv": {
       "version": "9",
       "source": "http://aber-owl.net/media/ontologies/MMUSDV",
-      "filetype": "owl"
+      "filename": "mmusdv.owl"
     },
     "NCBITaxon": {
       "version": "v2023-06-20",
       "source": "https://github.com/obophenotype/ncbitaxon/releases/download",
-      "filetype": "owl.gz"
+      "filename": "ncbitaxon.owl.gz"
     },
     "UBERON": {
       "version": "v2024-01-18",
       "source": "https://github.com/obophenotype/uberon/releases/download",
-      "filetype": "owl"
+      "filename": "uberon.owl"
     },
     "PATO": {
       "version": "v2023-05-18",
       "source": "https://github.com/pato-ontology/pato/raw",
-      "filetype": "owl"
+      "filename": "pato.owl"
     }
   }
 }
diff --git a/tools/ontology-builder/src/all_ontology_generator.py b/tools/ontology-builder/src/all_ontology_generator.py
@@ -62,7 +62,7 @@ def download(_ontology: str, _url: str) -> None:
 
     def _build_url(_ontology: str) -> str:
         onto_ref_data = ontology_info[_ontology]
-        return f"{onto_ref_data['source']}/{onto_ref_data['version']}/{ontology.lower()}.{onto_ref_data['filetype']}"
+        return f"{onto_ref_data['source']}/{onto_ref_data['version']}/{onto_ref_data['filename']}"
 
     threads = []
     for ontology, _ in ontology_info.items():

diff --git a/tools/ontology-builder/tests/test_all_ontology_generator.py b/tools/ontology-builder/tests/test_all_ontology_generator.py
@@ -13,7 +13,7 @@ def mock_ontology_info():
         "ontology_name": {
             "source": "http://example.com",
             "version": "v1",
-            "filetype": "owl",
+            "filename": "ontology_name.owl",
         }
     }
 
@@ -50,7 +50,9 @@ def test_get_ontology_info_file_default(mock_ontology_info_file):
     ontology_info = _get_ontology_info_file(ontology_info_file=mock_ontology_info_file)
 
     # Assertion
-    assert ontology_info == {"ontology_name": {"source": "http://example.com", "version": "v1", "filetype": "owl"}}
+    assert ontology_info == {
+        "ontology_name": {"source": "http://example.com", "version": "v1", "filename": "ontology_name.owl"}
+    }
 
 
 def test_get_ontology_info_file_version(mock_ontology_info_file):