search

NCATSTranslator / Feedback

A repo for tracking gaps in Translator data and finding ways to fill them.
7 stars 0 forks source link

UI: Some results are missing evidence #164

Closed khanspers closed 1 year ago

khanspers commented 1 year ago

Query: What may treat Duchenne Muscular Dystrophy? (run in https://ui.transltr.io/) PK: c0b054c9-381d-48a6-bfff-b534821f478f

7 of the top 20 results have no evidence given for the treats edge. Tracking down the top 4 results without evidence in ARAX, it looks like the only ara that's returning all 4 is improving. BTE also returns two of them (Drisapersen and Domagruzomab), and provides PMIDs for the edge, but this is not visible in the UI.

Screen Shot 2023-04-25 at 2 57 25 PM

sierra-moxon commented 1 year ago

from TAQA: Andy: yes they provide PMIDs, but they might not be using "biolink:publications" instead using "biolink:Publication" -- tagged with the wrong biolink attribute. Gus: node binding subclass issue - not picking up certain diseases with publications associated with the edge. Might be two or three reasons.

gprice1129 commented 1 year ago

Looked into this issue and found the following (link to query):

Drisapersen The evidence issue turns out to be that the CURIEs returned for Drisapersen by Improving Agent and BTE are different: Improving Agent: CHEMBL.COMPOUND:CHEMBL2108735 BTE: UMLS:C3886496

If you search for Drisapersen in the text search you can see two results with the same label. The second one has the included evidence from BTE.

Domagrozumab Same as for Drisapersen: Improving Agent: CHEMBL.COMPOUND:CHEMBL3544991 BTE: UMLS:C5139888

(8S,10S,13S,14S,16R,17R)... The underlying CURIE is PUBCHEM.COMPOUND:122197544. The CURIE only occurs in Improving Agent's answer and doesn't have any publications attached. The UI display that will be released next week should at least show that there is one piece of evidence which is the source (chembl).

Ramatercept Same issue as for (8S,10S,13S,14S,16R,17R)... Underlying CURIE is CHEMBL.COMPOUND:CHEMBL2108674

Going to include @cbizon and @gaurav to look into possible resolver issue.

gaurav commented 1 year ago

Thanks for reporting this! We'll look into ways of improving our CHEMBL.COMPOUND connections at https://github.com/TranslatorSRI/Babel/issues/131

Genomewide commented 1 year ago

This appears to be an old issue that was due to the fact that we only showed publications as evidence. is resolved now.