add explicit OLS search before querying Zooma

cmat/trait_mapping/main.py

@@ -4,10 +4,11 @@
 from collections import Counter
 
 from cmat.clinvar_xml_io import ClinVarTrait
+from cmat.trait_mapping.ols import get_uri_from_exact_match
 from cmat.trait_mapping.output import output_trait
 from cmat.trait_mapping.oxo import get_oxo_results
 from cmat.trait_mapping.oxo import uris_to_oxo_format
-from cmat.trait_mapping.trait import Trait
+from cmat.trait_mapping.trait import Trait, OntologyEntry
 from cmat.trait_mapping.trait_names_parsing import parse_trait_names
 from cmat.trait_mapping.zooma import get_zooma_results
 
@@ -32,8 +33,9 @@ def get_uris_for_oxo(zooma_result_list: list) -> set:
 
 def process_trait(trait: Trait, filters: dict, zooma_host: str, oxo_target_list: list, oxo_distance: int, target_ontology: str = 'EFO') -> Trait:
     """
-    Process a single trait. Find any mappings in Zooma. If there are no high confidence Zooma
-    mappings that are in EFO then query OxO with any high confidence mappings not in EFO.
+    Process a single trait. First look for an exact string match in the target ontology and return immediately if found.
+    Otherwise find any mappings in Zooma. If there are no high confidence Zooma mappings that are in EFO then query OxO
+    with any high confidence mappings not in EFO.
 
     :param trait: The trait to be processed.
     :param filters: A dictionary of filters to use for querying Zooma.
@@ -47,6 +49,11 @@ def process_trait(trait: Trait, filters: dict, zooma_host: str, oxo_target_list:
     """
     logger.debug('Processing trait {}'.format(trait.name))
 
+    string_match_uri = get_uri_from_exact_match(trait.name.lower(), target_ontology)
+    if string_match_uri:
+        trait.finished_mapping_set.add(OntologyEntry(string_match_uri, trait.name.lower()))
+        return trait
+
     trait.zooma_result_list = get_zooma_results(trait.name.lower(), filters, zooma_host, target_ontology)
     trait.process_zooma_results()
     if (trait.is_finished

cmat/trait_mapping/ols.py

@@ -1,3 +1,4 @@
+import os
 from functools import lru_cache
 import logging
 import requests
@@ -121,3 +122,31 @@ def get_replacement_term(uri: str, ontology: str = 'EFO') -> str:
     if response_json["term_replaced_by"] is not None:
         return response_json["term_replaced_by"]
     return ""
+
+
+@lru_cache(maxsize=16384)
+@retry(exceptions=(ConnectionError, requests.RequestException), tries=4, delay=2, backoff=1.2, jitter=(1, 3))
+def get_uri_from_exact_match(text, ontology='EFO'):
+    """
+    Finds URI from target ontology for a given text based on exact string match.
+
+    :param text: String to search for
+    :param ontology: ID of target ontology to query (default EFO)
+    :return: URI of matching term or None if not found
+    """
+    search_url = os.path.join(OLS_SERVER, f'api/search?ontology={ontology}&q={text}&queryFields=label&exact=true')
+    response = requests.get(search_url)
+    response.raise_for_status()
+    data = response.json()
+    if 'response' in data:
+        results = data['response']['docs']
+        candidates = set()
+        for result in results:
+            # Check that we've found the term exactly (strict case-insensitive string match)
+            if result['label'].lower() == text.lower():
+                candidates.add(result['iri'])
+        # Only return a result if we can find it unambiguously
+        if len(candidates) == 1:
+            return candidates.pop()
+    logger.warning(f'Could not find an IRI for {text}')
+    return None

tests/trait_mapping/test_main.py

@@ -68,10 +68,10 @@ def test_main():
 def test_process_trait_exact_match():
     # Exact match with MONDO:0009061 (in EFO and Mondo)
     trait_name = 'Cystic Fibrosis'
-    # Don't use any data sources in Zooma as those will come back as high-confidence matches
-    zooma_filters = {'ontologies': 'efo,mondo,hp',
-                     'required': 'none',
-                     'preferred': 'none'}
+    # Use our default Zooma filters
+    zooma_filters = {'ontologies': 'efo,ordo,hp,mondo',
+                     'required': 'cttv,eva-clinvar,clinvar-xrefs,gwas',
+                     'preferred': 'eva-clinvar,cttv,gwas,clinvar-xrefs'}
     zooma_host = 'https://www.ebi.ac.uk'
     # Don't use OxO
     oxo_targets = []