Chem Entity causes increase of Gene that causes target Disease - why is this returned for MVP1 - what drugs may treat?

[sstemann]

Steps:

1. CI UI, What drugs may treat dentin dysplasia: https://ui.ci.transltr.io/results?q=72a5d7c4-c88c-4fb9-936c-b2a89f885b66
2. Expand the fourth result (scored 98.1)

3. in ARAX UI it looks like its coming from BTE: https://arax.ncats.io/?r=72a5d7c4-c88c-4fb9-936c-b2a89f885b66
   

[andrewsu]

Apologies, knowing nothing about Anti-Mullerian Hormone or dentin dysplasia, why is this a problematic answer?

[sierra-moxon]

I swear I had added a screenshot to this while doing the triage because I had the same question. Ah well, here it is. :) I came to the conclusion that it might be because the qualifiers indicate that this chemical increases the abundance of the gene, and the gene has a "causes" relationship with the disease. As a naive user, if a gene (or gene product) causes a disease, then more of the gene (or gene product) would be the opposite desired effect of a chemical treatment? I couldn't tell from the "causes" edge metadata between the gene and the disease if "causes" was from an abundance or deficit of the gene (or gene product).

[andrewsu]

Thanks @sierra-moxon. Yes, that makes sense. But I think because, as you noted, the nature of the relationship between the gene and the disease isn't specified more concretely, so I think this is a result that we'd still want to show. (And it's actually a design principle in BTE that directionality is a secondary consideration. i.e., I'd much rather show an answer that reflects a biologically correct relationship in the "wrong" direction than not show the relationship at all. Directionality can be adjusted in many cases, and that's something I'd rather the user / SME evaluate anyway.)

Unless there is another concern here that I'm missing, I suggest closing this issue.

[sierra-moxon]

the only other two things I could find were that:

ACVR1 activity/expression - causes -> dental dysplasia was assessed in mice (Acvr1) (https://pubmed.ncbi.nlm.nih.gov/30519900/ - SEMMEDDB)

• found curated annotations in MGI, with many more details about the genotype that might prevent curators at MGI from declaring Acvr1 causative? (not sure, SME expertise would be very valuable here).

MIS - affects (increases abundance of) -> ACVR1 was assessed in rats (https://pubmed.ncbi.nlm.nih.gov/16687449/ - SEMMEDDB)

• found curated annotations for this paper in RGD, but they don't reference ACVR1

[sstemann]

@sierra-moxon does this change anything? in the same result set i see both increase and decrease:

[sierra-moxon]

I did reach out to Monarch Initiative for clarification here, and we found that there are assertions in the SRI Reference KG between Acvr1 and "abnormal mandible morphology" (from a different pub). We don't see the "dental dysplasia" assertion directly in a curated resource, but in my opinion, SEMMEDDB is adding value here ahead of curation.

Some things that would have helped this investigation go more quickly and result in more confidence:

• "knock out" designation on the gene would be helpful
• "taxon" metadata would be helpful
• Link the gene itself to the appropriate external resource to tell me more about its phenotypes (this is already in discussion via other tickets in this repo: include link outs to nodes based on their identifiers in UI results #91)
• Be able to see all the publications in Translator that have Acvr1 "mouth" phenotype assertions (something like a link to "related search results" or "did you mean?" functionality)

This is probably a separate set of findings from those in Sarah's original post, which she has clarified here. I'll put this in "Review at TAQA" for now? :)

[sierra-moxon]

we can use this ticket to look at attributes from mygene.info and use it to retrieve the taxon

[sierra-moxon]

I think the summary of this ticket is that we won't do anything to change the results.
Should we close?