Add explicit typedefs for GeoNames and ICD10 (#240)

biopragmatics · Nov 23, 2024 · fc8f639 · fc8f639
1 parent df343f4
commit fc8f639
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 8 deletions.
diff --git a/src/pyobo/sources/geonames.py b/src/pyobo/sources/geonames.py
@@ -9,7 +9,7 @@
 from pystow.utils import read_zipfile_csv
 from tqdm import tqdm
 
-from pyobo import Obo, Term
+from pyobo import Obo, Term, TypeDef
 from pyobo.struct import Reference, part_of
 from pyobo.utils.path import ensure_df, ensure_path
 
@@ -22,14 +22,15 @@
 ADMIN1_URL = "https://download.geonames.org/export/dump/admin1CodesASCII.txt"
 ADMIN2_URL = "https://download.geonames.org/export/dump/admin2Codes.txt"
 CITIES_URL = "https://download.geonames.org/export/dump/cities15000.zip"
+CODE_TYPEDEF = TypeDef.default(PREFIX, "code")
 
 
 class GeonamesGetter(Obo):
     """An ontology representation of GeoNames."""
 
     ontology = PREFIX
     dynamic_version = True
-    typedefs = [part_of]
+    typedefs = [part_of, CODE_TYPEDEF]
 
     def iter_terms(self, force: bool = False) -> Iterable[Term]:
         """Iterate over terms in the ontology."""
@@ -80,7 +81,7 @@ def get_code_to_country(*, force: bool = False) -> Mapping[str, Term]:
             term.append_synonym(fips)
         if pd.notna(iso3):
             term.append_synonym(iso3)
-        term.annotate_literal("code", code)
+        term.annotate_literal(CODE_TYPEDEF, code)
         code_to_country[code] = term
     logger.info(f"got {len(code_to_country):,} country records")
     return code_to_country
@@ -107,7 +108,7 @@ def get_code_to_admin1(
         term = Term.from_triple(
             "geonames", identifier, name if pd.notna(name) else None, type="Instance"
         )
-        term.annotate_literal("code", code)
+        term.annotate_literal(CODE_TYPEDEF, code)
         code_to_admin1[code] = term
 
         country_code = code.split(".")[0]
@@ -135,7 +136,7 @@ def get_code_to_admin2(
         term = Term.from_triple(
             "geonames", identifier, name if pd.notna(name) else None, type="Instance"
         )
-        term.annotate_literal("code", code)
+        term.annotate_literal(CODE_TYPEDEF, code)
         code_to_admin2[code] = term
         admin1_code = code.rsplit(".", 1)[0]
         admin1_term = code_to_admin1.get(admin1_code)

diff --git a/src/pyobo/sources/icd10.py b/src/pyobo/sources/icd10.py
@@ -17,7 +17,7 @@
     get_icd,
     visiter,
 )
-from ..struct import Obo, Reference, Synonym, Term
+from ..struct import Obo, Reference, Synonym, Term, has_category
 from ..utils.path import prefix_directory_join
 
 __all__ = [
@@ -35,6 +35,7 @@ class ICD10Getter(Obo):
 
     ontology = PREFIX
     dynamic_version = True
+    typedefs = [has_category]
 
     def iter_terms(self, force: bool = False) -> Iterable[Term]:
         """Iterate over terms in the ontology."""
@@ -81,8 +82,7 @@ def _extract_icd10(res_json: Mapping[str, Any]) -> Term:
         synonyms=synonyms,
         parents=parents,
     )
-
-    rv.annotate_literal("class_kind", res_json["classKind"])
+    rv.annotate_literal(has_category, res_json["classKind"])
 
     return rv
 

diff --git a/src/pyobo/struct/__init__.py b/src/pyobo/struct/__init__.py
@@ -17,6 +17,7 @@
     enables,
     from_species,
     gene_product_member_of,
+    has_category,
     has_gene_product,
     has_member,
     has_part,
@@ -47,6 +48,7 @@
     "enables",
     "from_species",
     "gene_product_member_of",
+    "has_category",
     "has_gene_product",
     "has_member",
     "has_part",