litcoin name res without UMLS but with autocomplete

RobokopU24 · Feb 6, 2024 · dfc8f61 · dfc8f61
1 parent 0871332
commit dfc8f61
Show file tree

Hide file tree

Showing 4 changed files with 12 additions and 13 deletions.
diff --git a/Common/data_sources.py b/Common/data_sources.py
@@ -18,7 +18,7 @@
 HMDB = 'HMDB'
 HUMAN_GOA = 'HumanGOA'
 INTACT = 'IntAct'
-LITCOIN = 'LitCoin_without_umls'
+LITCOIN = 'LitCoin_without_umls_with_autocomplete'
 LITCOIN_SAPBERT = 'LitCoinSapBERT_without_umls'
 MONARCH_KG = 'MonarchKG'
 MONDO_PROPS = 'MONDOProps'

diff --git a/Common/normalization.py b/Common/normalization.py
@@ -68,7 +68,6 @@ def hit_node_norm_service(self, curies, retries=0):
         resp: requests.models.Response = requests.post(f'{self.node_norm_endpoint}get_normalized_nodes',
                                                        json={'curies': curies,
                                                              'conflate': self.conflate_node_types,
-                                                             'drug_chemical_conflate': self.conflate_node_types,
                                                              'description': True})
         if resp.status_code == 200:
             # if successful return the json as an object
@@ -535,16 +534,14 @@ def call_name_resolution(name: str, biolink_type: str, retries=0, logger=None):
     nameres_payload = {
         "string": name,
         "biolink_type": biolink_type if biolink_type else "",
-        "autocomplete": False
+        "autocomplete": True,
+        "exclude_prefixes": "UMLS"
     }
     nameres_result = requests.get(NAME_RESOLVER_URL, params=nameres_payload, headers=NAME_RESOLVER_HEADERS)
     if nameres_result.status_code == 200:
         # return the first result if there is one
         nameres_json = nameres_result.json()
-        if nameres_json:
-            for result in nameres_json:
-                if not result['curie'].startswith('UMLS'):
-                    return result
+        return nameres_json[0] if nameres_json else None
     else:
         error_message = f'Non-200 result from name resolution ({NAME_RESOLVER_URL}): ' \
                         f'Status {nameres_result.status_code} - {nameres_payload}.'

diff --git a/graph_specs/litcoin-graph-spec.yml b/graph_specs/litcoin-graph-spec.yml
@@ -9,13 +9,13 @@ graphs:
     sources:
       - source_id: LitCoinSapBERT_without_umls
 
-  - graph_id: LitCoin_without_umls
-    graph_name: LitCoin_without_umls
+  - graph_id: LitCoin_without_umls_with_autocomplete
+    graph_name: LitCoin_without_umls_with_autocomplete
     graph_description: A graph for the LitCoin project using Name Resolver for name resolution without umls identifiers
     conflation: True
     strict_normalization: True
-    output_format: neo4j
+    output_format: None
     sources:
-      - source_id: LitCoin_without_umls
+      - source_id: LitCoin_without_umls_with_autocomplete
 
 
diff --git a/parsers/LitCoin/src/loadLitCoin.py b/parsers/LitCoin/src/loadLitCoin.py
@@ -71,7 +71,7 @@
 ##############
 class LitCoinLoader(SourceDataLoader):
 
-    source_id: str = 'LitCoin_without_umls'
+    source_id: str = 'LitCoin_without_umls_with_autocomplete'
     provenance_id: str = 'infores:robokop-kg'  # TODO - change this to a LitCoin infores when it exists
     parsing_version: str = '1.4'
 
@@ -250,7 +250,9 @@ def parse_llm_output(self, llm_output):
         return valid_responses
 
     def name_resolution_function(self, node_name, preferred_biolink_node_type, retries=0):
-        return call_name_resolution(node_name, preferred_biolink_node_type, retries, logger=self.logger)
+        return call_name_resolution(node_name,
+                                    preferred_biolink_node_type,
+                                    retries, logger=self.logger)
 
     def standardize_name_resolution_results(self, name_res_json):
         if not name_res_json: