Merge pull request #20 from microbiomedata/issue-12-prefix-casing

Match ontology CURIE prefix case-insensitively (closes #12)

Author: sierra-moxon

src/ontology_loader/ontology_processor.py

@@ -50,6 +50,9 @@ def __init__(self, ontology: str):
 
         """
         self.ontology = ontology
+        # Cache the lowercased ontology name once; `_matches_ontology` is called
+        # in hot loops over millions of entities/ancestors at NCBITaxon scale.
+        self._ontology_lc = ontology.lower()
         self.ontology_db_path = self.download_and_prepare_ontology()
         self.adapter = get_adapter(f"sqlite:{self.ontology_db_path}")
         self.adapter.precompute_lookups()  # Optimize lookups
@@ -115,20 +118,24 @@ def _create_ontology_class(self, entity_id, is_obsolete=False):
 
         return ontology_class
 
+    def _matches_ontology(self, entity_id: str) -> bool:
+        """Case-insensitive check that ``entity_id`` is a CURIE in this ontology."""
+        head, sep, _ = entity_id.partition(":")
+        return bool(sep) and head.lower() == self._ontology_lc
+
     def get_terms_and_metadata(self):
-        """Retrieve all terms that start with the ontology prefix and return a list of OntologyClass objects."""
+        """Retrieve all terms that belong to this ontology and return a list of OntologyClass objects."""
         ontology_classes = []
-        ontology_prefix = self.ontology.upper() + ":"
 
         # Process non-obsolete entities
         for entity in self.adapter.entities(filter_obsoletes=True):
-            if entity.startswith(ontology_prefix):
+            if self._matches_ontology(entity):
                 ontology_class = self._create_ontology_class(entity, is_obsolete=False)
                 ontology_classes.append(ontology_class)
 
         # Process obsolete entities
         for obsolete_entity in self.adapter.obsoletes():
-            if obsolete_entity.startswith(ontology_prefix):
+            if self._matches_ontology(obsolete_entity):
                 ontology_class = self._create_ontology_class(obsolete_entity, is_obsolete=True)
                 ontology_classes.append(ontology_class)
 
@@ -143,15 +150,14 @@ def get_relations_closure(self, predicates=None, ontology_terms: list = None) ->
         :return: Tuple of (ontology_relations, updated_ontology_terms)
         """
         predicates = ["rdfs:subClassOf", "BFO:0000050"] if predicates is None else predicates
-        ontology_prefix = self.ontology.upper() + ":"
         ontology_relations = []
 
         # Create dictionary for fast lookup of ontology terms
         ontology_terms_dict = {term.id: term for term in (ontology_terms or [])}
 
         # Get all relevant entities in one pass
         logger.info("Collecting relevant entities...")
-        relevant_entities = set(entity for entity in self.adapter.entities() if entity.startswith(ontology_prefix))
+        relevant_entities = set(entity for entity in self.adapter.entities() if self._matches_ontology(entity))
         logger.info(f"Found {len(relevant_entities)} relevant entities")
 
         # Process all direct relationships in one batch
@@ -177,7 +183,7 @@ def get_relations_closure(self, predicates=None, ontology_terms: list = None) ->
             ancestors = set(
                 ancestor
                 for ancestor in self.adapter.ancestors(entity, reflexive=True, predicates=predicates)
-                if ancestor.startswith(ontology_prefix)
+                if self._matches_ontology(ancestor)
             )
 
             # Create relations for each ancestor

tests/test_ontology_processor.py

@@ -1,8 +1,45 @@
 """Test OntologyProcessor class and its methods."""
 
+import pytest
+
 from src.ontology_loader.ontology_processor import OntologyProcessor
 
 
+@pytest.mark.parametrize(
+    "ontology_name, entity_id, expected",
+    [
+        # Same-case prefixes (the historical path)
+        ("envo", "ENVO:00002005", True),
+        ("envo", "envo:00002005", True),
+        ("uberon", "UBERON:0000001", True),
+        ("po", "PO:0000001", True),
+        # Mixed-case prefixes that the prior `.upper()`-based filter dropped silently
+        ("ncbitaxon", "NCBITaxon:9606", True),
+        ("ncbitaxon", "NCBITAXON:9606", True),
+        ("ncbitaxon", "ncbitaxon:9606", True),
+        ("chebi", "CHEBI:12345", True),
+        # Wrong ontology — must reject
+        ("envo", "UBERON:0000001", False),
+        ("ncbitaxon", "PR:Q9606", False),
+        # Missing colon — must reject
+        ("ncbitaxon", "NCBITaxon", False),
+        ("envo", "ENVO", False),
+        ("envo", "", False),
+    ],
+)
+def test_matches_ontology(ontology_name, entity_id, expected):
+    """`_matches_ontology` compares the CURIE head case-insensitively to the configured ontology."""
+
+    # Avoid the heavy `OntologyProcessor.__init__` (which downloads + opens sqlite); the method
+    # only depends on `self._ontology_lc`, so a minimal stand-in object is sufficient.
+    class _Fake:
+        pass
+
+    fake = _Fake()
+    fake._ontology_lc = ontology_name.lower()
+    assert OntologyProcessor._matches_ontology(fake, entity_id) is expected
+
+
 def test_ontology_processor():
     """Test OntologyProcessor initialization and ontology retrieval."""
     ontology_name = "envo"