NCATSTranslator / Feedback

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

Two paths because one predicate is "has phenotype" and another predicate is "phenotype of" #965

Open sstemann opened 1 week ago

sstemann commented 1 week ago

these paths have identical EPC but looks like we need two because the predicates are different? anyway to consolidate here?

https://ui.test.transltr.io/results?l=Postaxial%20Acrofacial%20Dysostosis&i=MONDO:0009903&t=0&r=0&q=019c52a4-065a-4f6d-a99e-d2f99fc1b963

image
dnsmith124 commented 4 days ago

I'm not sure where responsibility for this should lie. On the one hand the UI is sort of the "bottom of the hill" as it were, so it makes sense that we might handle some sort of consolidation in situations like this. On the other hand that sounds a bit like we'd be approaching reasoning territory.

andrewsu commented 4 days ago

Hmm, I'm a bit confused on this one. From https://arax.ci.transltr.io/?r=019c52a4-065a-4f6d-a99e-d2f99fc1b963, I see the BTE PK is 45865aa0-6cf9-4ec3-a2b7-2c69a97216ff, which I downloaded from https://ars.test.transltr.io/ars/api/messages/45865aa0-6cf9-4ec3-a2b7-2c69a97216ff. I see 52 edges of biolink:has_phenotype, but zero that use biolink:phenotype_of.

$ python3 -m json.tool 45865aa0-6cf9-4ec3-a2b7-2c69a97216ff.json  | grep -c 'biolink:has_phenotype'
52
$ python3 -m json.tool 45865aa0-6cf9-4ec3-a2b7-2c69a97216ff.json  | grep -c 'biolink:phenotype_of'
0

So I don't think BTE is reporting those edges. @dnsmith124 is there an easy way to extract the edge IDs for those edges that the UI is seeing? Might help investigating further...

dnsmith124 commented 4 days ago

Looking at the BTE response it ARAX I'm fairly certain that the phenotype_of edges are the result of the UI flipping those predicates in order to make the paths linear from the subject node to the object node.

@gprice1129 is there a duplication issue here on the UI's end or are the resulting paths in the above result correct?