Update HGNC export (#486)

Motivated by biopragmatics/bioregistry#1811

This PR does the following:

1. Adds missing relations (e.g., ones that appear in logical definition
of other ones)
2. Adds explicit labels for placeholder SO terms that haven't been
reviewed yet (see
The-Sequence-Ontology/SO-Ontologies#667 and
The-Sequence-Ontology/SO-Ontologies#668)
3. Directly incorporate HGNC gene groups content into HGNC export (while
simultaneously addressing some version management issues for HGNC gene
groups)

This does not solve the more generic issue that PyOBO exports OFN and
OWL products using IRIs instead of stringified CURIEs in annotations.
This means that if the IRI isn't defined, then Protege tries its best to
guess a label, and which pollutes the class list when exploring.

- protegeproject/protege#1157
- protegeproject/protege#1321
- protegeproject/protege#1322
- protegeproject/protege#1323

Author: cthoyt

src/pyobo/resources/so.py

@@ -21,6 +21,12 @@
 
 def get_so_name(so_id: str) -> str | None:
     """Get the name from the identifier."""
+    if so_id == "0003002":
+        # see https://github.com/The-Sequence-Ontology/SO-Ontologies/pull/668
+        return "viral integration site"
+    if so_id == "0003001":
+        # see https://github.com/The-Sequence-Ontology/SO-Ontologies/pull/667
+        return "cluster RNA gene"
     return load_so().get(so_id)
 
 

src/pyobo/sources/hgnc/hgnc.py

@@ -14,27 +14,25 @@
 
 from pyobo.api.utils import get_version
 from pyobo.resources.so import get_so_name
-from pyobo.struct import (
-    Annotation,
-    Obo,
-    OBOLiteral,
-    Reference,
-    Term,
-    from_species,
-    has_gene_product,
-    is_mentioned_by,
-    member_of,
-    orthologous,
-    transcribes_to,
-)
+from pyobo.sources.hgnc.hgncgenefamily import GENE_GROUP_PREFIX, get_gene_family_terms
+from pyobo.struct import Annotation, Obo, OBOLiteral, Reference, Term
 from pyobo.struct.struct import gene_symbol_synonym, previous_gene_symbol, previous_name
 from pyobo.struct.typedef import (
     comment,
+    enables,
     ends,
     exact_match,
+    from_species,
     gene_product_enables,
+    gene_product_of,
+    has_gene_product,
+    has_member,
+    is_mentioned_by,
     located_in,
+    member_of,
+    orthologous,
     starts,
+    transcribes_to,
 )
 from pyobo.utils.path import ensure_path
 
@@ -137,6 +135,7 @@
     "homeodb",  # TODO add to bioregistry, though this is defunct
     "mamit-trnadb",  # TODO add to bioregistry, though this is defunct
     "mane_select",  # TODO
+    "gene_group",  # gene_group_id is needed, but this just has label
 }
 
 #: A mapping from HGNC's locus_type annotations to sequence ontology identifiers
@@ -195,8 +194,10 @@ class HGNCGetter(Obo):
     typedefs = [
         from_species,
         has_gene_product,
+        gene_product_of,
         gene_product_enables,
         transcribes_to,
+        has_member,
         orthologous,
         member_of,
         exact_match,
@@ -205,6 +206,7 @@ class HGNCGetter(Obo):
         starts,
         ends,
         comment,
+        enables,
     ]
     synonym_typedefs = [
         previous_name,
@@ -262,6 +264,8 @@ def get_terms(version: str | None = None, force: bool = False) -> Iterable[Term]
         for so_id in sorted(_so_ids)
     ]
 
+    yield from get_gene_family_terms(version=version, force=force)
+
     statuses = set()
     for entry in tqdm(entries, desc=f"Mapping {PREFIX}", unit="gene", unit_scale=True):
         name, symbol, identifier = (
@@ -393,15 +397,10 @@ def get_terms(version: str | None = None, force: bool = False) -> Iterable[Term]
             term.append_mentioned_by(Reference(prefix="pubmed", identifier=str(pubmed_id)))
 
         gene_group_ids = entry.pop("gene_group_id", [])
-        gene_groups = entry.pop("gene_group", [])
-        for gene_group_id, gene_group_label in zip(gene_group_ids, gene_groups, strict=False):
+        for gene_group_id in gene_group_ids:
             term.append_relationship(
                 member_of,
-                Reference(
-                    prefix="hgnc.genegroup",
-                    identifier=str(gene_group_id),
-                    name=gene_group_label,
-                ),
+                Reference(prefix=GENE_GROUP_PREFIX, identifier=str(gene_group_id)),
             )
 
         for alias_symbol in entry.pop("alias_symbol", []):

src/pyobo/sources/hgnc/hgncgenefamily.py

@@ -5,16 +5,20 @@
 
 import pandas as pd
 
-from ...struct import Obo, Reference, Term, is_mentioned_by
+from ...api.utils import get_version
+from ...struct.struct import Obo, Reference, Term
 from ...struct.struct import abbreviation as symbol_type
-from ...struct.typedef import enables, exact_match, from_species
+from ...struct.typedef import enables, exact_match, from_species, is_mentioned_by
 from ...utils.path import ensure_df
 
 __all__ = [
+    "GENE_GROUP_REFERENCE",
+    "GENE_GROUP_TERM",
     "HGNCGroupGetter",
+    "get_gene_family_terms",
 ]
 
-PREFIX = "hgnc.genegroup"
+GENE_GROUP_PREFIX = "hgnc.genegroup"
 FAMILIES_URL = "https://storage.googleapis.com/public-download-files/hgnc/csv/csv/genefamily_db_tables/family.csv"
 FAMILIES_ALIAS_URL = "https://storage.googleapis.com/public-download-files/hgnc/csv/csv/genefamily_db_tables/family_alias.csv"
 HIERARCHY_URL = "https://storage.googleapis.com/public-download-files/hgnc/csv/csv/genefamily_db_tables/hierarchy.csv"
@@ -23,19 +27,21 @@
 class HGNCGroupGetter(Obo):
     """An ontology representation of HGNC's gene group nomenclature."""
 
-    ontology = PREFIX
+    ontology = GENE_GROUP_PREFIX
     bioversions_key = "hgnc"
     synonym_typedefs = [symbol_type]
     typedefs = [from_species, enables, exact_match, is_mentioned_by]
 
     def iter_terms(self, force: bool = False) -> Iterable[Term]:
         """Iterate over terms in the ontology."""
-        return get_terms(force=force)
+        return get_gene_family_terms(force=force)
 
 
-def get_hierarchy(force: bool = False) -> Mapping[str, list[str]]:
+def get_hierarchy(*, version: str | None = None, force: bool = False) -> Mapping[str, list[str]]:
     """Get the HGNC Gene Families hierarchy as a dictionary."""
-    df = ensure_df(PREFIX, url=HIERARCHY_URL, force=force, sep=",")
+    if version is None:
+        version = get_version("hgnc")
+    df = ensure_df(GENE_GROUP_PREFIX, url=HIERARCHY_URL, force=force, sep=",", version=version)
     d = defaultdict(list)
     for parent_id, child_id in df.values:
         d[child_id].append(parent_id)
@@ -44,51 +50,51 @@ def get_hierarchy(force: bool = False) -> Mapping[str, list[str]]:
 
 COLUMNS = ["id", "abbreviation", "name", "pubmed_ids", "desc_comment", "desc_go"]
 
+GENE_GROUP_REFERENCE = Reference(prefix="SO", identifier="0005855", name="gene group")
+GENE_GROUP_TERM = Term(reference=GENE_GROUP_REFERENCE)
 
-def get_terms(force: bool = False) -> Iterable[Term]:
+
+def get_gene_family_terms(*, version: str | None = None, force: bool = False) -> Iterable[Term]:
     """Get the HGNC Gene Group terms."""
-    terms = list(_get_terms_helper(force=force))
-    hierarchy = get_hierarchy(force=force)
+    if version is None:
+        version = get_version("hgnc")
+
+    terms = list(_get_terms_helper(force=force, version=version))
+    hierarchy = get_hierarchy(force=force, version=version)
 
-    id_to_term = {term.reference.identifier: term for term in terms}
+    id_to_term = {term.identifier: term for term in terms}
     for child_id, parent_ids in hierarchy.items():
         child: Term = id_to_term[child_id]
         for parent_id in parent_ids:
-            parent: Term = id_to_term[parent_id]
-            child.append_parent(
-                Reference(
-                    prefix=PREFIX,
-                    identifier=parent_id,
-                    name=parent.name,
-                )
-            )
-    gene_group = Reference(prefix="SO", identifier="0005855", name="gene group")
-    yield Term(reference=gene_group)
+            child.append_parent(id_to_term[parent_id])
+    yield GENE_GROUP_TERM
     for term in terms:
         if not term.parents:
-            term.append_parent(gene_group)
+            term.append_parent(GENE_GROUP_REFERENCE)
     yield from terms
 
 
-def _get_terms_helper(force: bool = False) -> Iterable[Term]:
-    alias_df = ensure_df(PREFIX, url=FAMILIES_ALIAS_URL, force=force, sep=",")
+def _get_terms_helper(version: str, force: bool = False) -> Iterable[Term]:
+    alias_df = ensure_df(
+        GENE_GROUP_PREFIX, url=FAMILIES_ALIAS_URL, force=force, sep=",", version=version
+    )
     aliases = defaultdict(set)
     for _id, family_id, alias in alias_df.values:
         aliases[family_id].add(alias)
 
-    df = ensure_df(PREFIX, url=FAMILIES_URL, force=force, sep=",")
+    df = ensure_df(GENE_GROUP_PREFIX, url=FAMILIES_URL, force=force, sep=",", version=version)
     for gene_group_id, symbol, name, pubmed_ids, definition, desc_go in df[COLUMNS].values:
         if not definition or pd.isna(definition):
             definition = None
         term = Term(
-            reference=Reference(prefix=PREFIX, identifier=gene_group_id, name=name),
+            reference=Reference(prefix=GENE_GROUP_PREFIX, identifier=gene_group_id, name=name),
             definition=definition,
         )
         if pubmed_ids and pd.notna(pubmed_ids):
-            for s in pubmed_ids.replace(" ", ",").split(","):
-                s = s.strip()
-                if s:
-                    term.append_mentioned_by(Reference(prefix="pubmed", identifier=s))
+            for pubmed_id in pubmed_ids.replace(" ", ",").split(","):
+                pubmed_id = pubmed_id.strip()
+                if pubmed_id:
+                    term.append_mentioned_by(Reference(prefix="pubmed", identifier=pubmed_id))
         if desc_go and pd.notna(desc_go):
             go_id = desc_go[len("http://purl.uniprot.org/go/") :]
             term.append_relationship(enables, Reference(prefix="GO", identifier=go_id))

src/pyobo/struct/typedef.py

@@ -55,6 +55,7 @@
     "is_antagonist_of",
     "is_defined_by",
     "is_inverse_agonist_of",
+    "is_mentioned_by",
     "located_in",
     "mapping_has_confidence",
     "mapping_has_justification",