Biolink API taxon filtering not behaving as I expected

[willfitzhugh]

Hi, I've been using the GO API that is published here - http://api.geneontology.org/api - but am having some basic issues with the results. Not sure if it's an issue with the software or a misunderstanding of how the API is supposed to work.

I am making a call to get the genes associated with a GO term:

http://api.geneontology.org/api/bioentity/function/GO%3A0002544/genes?rows=100&relationship_type=involved_in

This returns genes from all organisms but I am interested in only human genes so then I tried:

http://api.geneontology.org/api/bioentity/function/GO%3A0002544/genes?rows=100&taxon=NCBITaxon%3A9606&relationship_type=involved_in

Which appears to return exactly the same results. I would have thought this would return only human genes. Can you help me with this?

Thanks!

Will

[kltm]

@willfitzhugh

Unfortunately, you seem to have actually found a bug/oversight in the API. I have opened a ticket and we are tracking it here: monarch-initiative/biolink-api#365 .

[willfitzhugh]

Thanks for your response and for adding it to the queue to get fixed.

[suzialeksander]

Note to self: Just checked this search and it appears the fix hasn't come through yet.

[willfitzhugh]

Hi, just checking, any progress on this?

[suzialeksander]

@kltm, it looks like the fix is in Ontobio according to monarch-initiative/biolink-api#365 but I'm still seeing this issue. Is there a release or something that needs to update before we see the API working as intended?

[suzialeksander]

Update, we are still working on this. The issue has been narrowed down to a mismatch of fields between Monarch Solr and GOlr: object_taxon vs taxon. Looks like there's at least one more step needed to fix completely.

[suzialeksander]

This looks like it's still an active bug, and I have pinged the biolink team again

[rbeagrie]

I've also just spent an hour trying to work out why taxon filtering wasn't working as expected in the API and just found this bug report. Is there any update on fixing this issue?

[kltm]

@rbeagrie My apologies for the slow movement on this. We're currently refactoring the codebase for the API to make fixes easier and this has slowed down some rollout and fixes. I'll check in on the ticket on the biolink tracker (monarch-initiative/biolink-api#365 (comment)).
In the meantime, I might suggest working with the files directly or consider maybe the SPARQL endpoint (if the query is likely to run before it hits the internal timeout).

[sierra-moxon]

auto-close with my PR, apologies - this should stay open until we deploy the new API.