search

NCATSTranslator / testing

Materials and tools for testing Translator components
1 stars 9 forks source link

ChemicalSubstance (PUBCHEM.COMPOUND:40466924) - (causes, causes_adverse_event) - Disease #82

Open sstemann opened 3 years ago

sstemann commented 3 years ago

Query: enalaprilDisease.json PK: ff36cd35-2d82-4b62-b550-4e2a5d8f017a PUBCHEM.COMPOUND: 40466924

Results Tracking Sheet

image

CaseyTa commented 3 years ago

COHD and OpenPredict are behaving as expected. No results from OpenPredict, and COHD returns the message:

None of the predicates in ['biolink:causes', 'biolink:causes_adverse_event'] are supported by COHD.

andrewsu commented 3 years ago

The SmartAPI / BTE knowledge graph only has predicates for biolink:related_to, biolink:treats and biolink:contraindicated_for. (For reference, kludgey analysis of the query without specifying a predicate is below, showing the counts of the predicates we do get between PUBCHEM.COMPOUND:40466924 and Diseases...)

$ curl --silent --location --request POST "https://api.bte.ncats.io/v1/query" --header "Content-Type: application/json" --data-raw "{     \"message\": {         \"query_graph\": {             \"nodes\": {                 \"n0\": {                     \"ids\": [
                         \"PUBCHEM.COMPOUND:40466924\"                     ],                     \"categories\": [                         \"biolink:ChemicalSubstance\"                     ]                 },                 \"n1\": {                     \"categories\": [
         \"biolink:Disease\"                     ]                 }             },             \"edges\": {                 \"e0\": {
          \"subject\": \"n0\",                     \"object\": \"n1\"                 }             }         }     } }" | python3 -m json.tool | grep '"predicate"' | sort | uniq -c | sort -k1nr
    232                     "predicate": "biolink:related_to",
     27                     "predicate": "biolink:treats",
      3                     "predicate": "biolink:contraindicated_for",
edeutsch commented 3 years ago

Hi everyone, following up on "where did our edges come from?": at @webyrd @andrewsu

For example, ARAX returned: CHEMBL.COMPOUND:CHEMBL577 biolink:causes MONDO:0006945 from SEMMEDDB but these are synonymized CURIEs, not the originals.

The question was: what are the originals? This was not visible in our Response. Looking deeper, the answer seems to be: UMLS:C0014025 biolink:causes UMLS:C0035067 from SEMMEDDB

We will add a feature of ARAX and KG2 to display this information in the future.

rajshruti18 commented 3 years ago

CAM-KP does not have data on diseases, so 0 result is expected.

andrewsu commented 3 years ago

After the change in https://github.com/biothings/BioThings_Explorer_TRAPI/issues/217, BTE now returns results for the biolink:causes_adverse_event predicate in the original query https://arax.ncats.io/?source=ARS&id=509200a2-9e4d-490e-b93a-c67178b7b358

image