GO terms bug

[ValWood]

Some sessions have obsolete terms e.g.

https://curation.pombase.org/pombe/curs/760045d7bb2de1aa

but these are not reported in this file

https://curation.pombase.org/dumps/latest_build/logs/log.2024-09-30-21-55-49.curation-tool-data-load-output

I thought this is where they went?

[kimrutherford]

I don't know what's going on here. Is the term taxon restricted? It's not obsolete: https://www.ebi.ac.uk/QuickGO/term/GO:1903139

but these are not reported in this file

Because the term isn't obsolete, those annotations get loaded into Chado without a warning: https://www.pombase.org/term/GO:1903139 https://www.pombase.org/reference/PMID:24498240

It seems like Canto is the problem. I'm looking into it now.

[ValWood]

Yes I was just going to add that it isn't an obsolete term!

[kimrutherford]

I noticed that parent term GO:0000196 has weird taxon constraints. Does it make sense to have two "Only in Taxon" constraints like this?

We create a pombe-only OBO file for loading into Canto using OWLTools. When I remove that taxon constraint from the input file (go-plus.owl), and re-run OWLTools the missing term (GO:1903139) reappears. So the constraint does look like at least part of the problem.

Parent term: https://www.ebi.ac.uk/QuickGO/term/GO:0000196

[kimrutherford]

It looks like the "Viridiplantae or Bacteria or Euglenozoa" constraint was added in this PR:

• https://github.com/geneontology/go-ontology/pull/26802

[ValWood]

Yep that taxon constraint is incorrect. I will open a ticket

cc @pgaudet @balhoff