Handle output of new owltools version when computing closure

[kimrutherford]

In the latest version of owltools, "part_of" has changed to "BFO:0000050" in the relation column. Other relations have changed too. eg. "negatively_regulates" in now "RO:0002212".

This needs fixing in the loader code before we can upgrade. Upgrading isn't urgent though.

Refs geneontology/go-ontology#16776

[ValWood]

Is the priority of this still medium? how quick is it?

[kimrutherford]

I've just made the changes to support the new output style. In case the results will be slightly different from the current version, I won't check the changes in today. I'll test it in the nightly load over the weekend.

[kimrutherford]

I'll test it in the nightly load over the weekend.

I've checked in the changes to support the new OWLTools. Tomorrow I'll see if I can spot any major ontology problems.

[kimrutherford]

Tomorrow I'll see if I can spot any major ontology problems.

There's good news and, er, other news.

The good news is that it mostly looks fine. The other news is that I spotted an ontology closure problem that been around for a while.

In the main site now, sister chromatid segregation (GO:0000819) has 184 gene annotated according to this page: https://www.pombase.org/term/GO:0000819
but in yesterday's load it has 186: http://pombase-dev.bioinformatics.nz/term/GO:0000819 (that's temporarily yesterday's load)

The difference is that "meiosis I spindle assembly checkpoint (GO:1905318)" annotations appear on yesterday's "sister chromatid segregation (GO:0000819)" page, but not after updating OWLTools.

Here's the GO terms and relations that connect those terms. There might be other paths, I didn't check. Maybe the problem is caused by the "regulates" relation?

• meiosis I spindle assembly checkpoint (GO:1905318)
• is_a meiotic spindle assembly checkpoint (GO:0033316)
• is_a meiotic spindle checkpoint (GO:0044779)
• is_a negative regulation of metaphase/anaphase transition of meiotic cell cycle (GO:1902103)
• is_a regulation of metaphase/anaphase transition of meiotic cell cycle (GO:1902102)
• is_a regulation of metaphase/anaphase transition of cell cycle (GO:1902099)
• is_a regulation of sister chromatid segregation (GO:0033045)
• regulates sister chromatid segregation (GO:0000819)

I don't think the OWLTools update is causing this problem. I checked the Chado databases going back a few weeks and sometimes "meiosis I spindle assembly checkpoint (GO:1905318)" is a descendent of "sister chromatid segregation (GO:0000819)" page and sometimes not.

This seems similar to geneontology/go-ontology#16687

[ValWood]

Yes! This is the problem with owl tools that we tried to track down ages ago.
The number 'flip-flop' even when the data does not change.
We need @cmungall @balhoff to look at this.
So, this problem has existed for a while, independently of the update owltools update.

[ValWood]

I can only find this ticket
pombase/website#699

I know I opened some tickets on the GO tracker about this. I described it as "annotation flip-flopping". I can't find any of the tickets relating to this issue. The only other thing I remember is that the examples were from "cytoskeleton organization" and "conjugation with cellular fusion"

[mah11]

In our earlier encounters with this problem, we pinned it on owltools (tho apparently some part of it that's unaffected by the update). The most informative tickets aren't the ones on the GO ontology tracker; they're #678 on this tracker and the linked ticket on the owltools tracker.

[ValWood]

Thanks.
I knew I had written flip-flopping somewhere:
#678

[kimrutherford]

I've commented in the owltools issue about this: owlcollab/owltools#256

[kimrutherford]

I've commented in the owltools issue about this: owlcollab/owltools#256

I'll close this one since there's an owltools issue for the problem.