F- is not part of K+ - Githubissues

NCATSTranslator / Feedback

A repo for tracking gaps in Translator data and finding ways to fill them.

7 stars 0 forks source link

F- is not part of K+ #750

Closed TranslatorIssueCreator closed 3 months ago

TranslatorIssueCreator commented 7 months ago

Type: Bug Report

URL: https://ui.test.transltr.io/main/results?l=Potassium%20Ion&i=PUBCHEM.COMPOUND:813&t=3&r=0&q=66f83ee1-3054-4d7a-8e7b-35def1c1a267

ARS PK: 66f83ee1-3054-4d7a-8e7b-35def1c1a267

Steps to reproduce:

search for ALPL

Screenshots:

sandrine-muller-research commented 7 months ago

In the query "What gene may be upregulated by potassium ion?" we get : ALPL through the path : from ARAGORN I do not see how F- is part of K+, and more generally that an ion is part of another ion?

sandrine-muller-research commented 7 months ago

same issue with : CaCO3 part of K+ (CAPN1,SOD,CASP3 results)

sstemann commented 7 months ago

@cbizon @uhbrar

cbizon commented 7 months ago

The Flouride has part Potassium ion is coming from semmeddb. Looks like both the KG2 and service provider versions. It looks good in the sense that it's coming from 7 or so papers, so I think it's getting past some of the filters. But when you dig into it, it looks like the papers are all from the same group, and even have nearly consecutive PMIDs and repeat the same text. Somebody got a lot of pubs out of this toothpaste. Anyway, I'm not sure if there's a way to improve the semmed filters to cover a case like this?

The Calcium carbonate part of K+ are the same problem (and same papers I think).

@andrewsu @saramsey

mikebada commented 7 months ago

And it looks like the snippets are not the actual snippets from which the assertion was extracted, so it's hard to tell why it's extracting this. (I'm guessing SemMedDB doesn't provide the actual snippets?) But yeah, definitely wrong.

saramsey commented 6 months ago

I think this should be fixed in RTX-KG2 as of KG2.9.0c (Octopus release), where we are no longer using biolink:has_part for this, but instead, biolink:coexists_with.

andrewsu commented 3 months ago

I'm going to close this as currently not-able-to-reproduce. When I search for "potassium" in the UI, I only get one option:

That corresponds to CHEBI:26216 (potassium), not PUBCHEM.COMPOUND:813 (potassium ion) in the example above. (These are not equivalent according to nodenorm.) and I do not see ALPL among the results for "potassium": https://ui.test.transltr.io/main/results?l=Potassium&i=CHEBI:26216&t=3&r=0&q=ab59272c-b892-4677-a78a-9856b4fb9df0