Annotations from Noctua models not reaching Xenbase GPAD on snapshot.

[malcolmfisher103]

In the time between June 14th and June 22nd nearly 300 annotations have disappeared from the Xenbase GPAD file (http://snapshot.geneontology.org/products/annotations/noctua_xenbase.gpad.gz). The missing annotations can still be found in the latest release version (http://release.geneontology.org/2020-06-01/products/annotations/noctua_xenbase.gpad.gz). Please find attached a file with the annotations that have been lost (Missing.lines.txt).

Some of these may be the result of legitimate changes in models but in many cases entire models are no longer represented. I can still go to the models in Noctua and see the annotations in the 'Workbenches' menu's 'Annotation preview' or 'Export GPAD' option from the 'Model' menu, see this model for an example.

Have any extra QC checks been added in the last couple of weeks that might account for this? I am not seeing any obvious problem or commonalities with the models that have been lost.

[kltm]

@malcolmfisher103 To confirm: the models are still in Noctua (i.e. http://noctua.geneontology.org/editor/graph/gomodel:581e072c00000295 (noting the corrected domain)), but you are not seeing them in the output GPAD?

If so, and the models are making it to GitHub, this will need input from @ukemi and @balhoff , and possibly get moved to the minerva tracker.

[malcolmfisher103]

@kltm Yes, the models are in Noctua and in the turtle file on GitHub but not in the output GPAD.

[ukemi]

@kltm, this is correct. The models are there, but in the recent past something about the gpad output has changed and there are now missing annotations. The particular model above is a mixed species model, which may explain the problem. But, I don't think the one was that we looked at together this morning was it @malcolmfisher103?

[ukemi]

I just split out the human process from the frog process (GO:0001558).

[malcolmfisher103]

There are other purely frog examples such as 'gomodel:5a7e68a100001201', the one we looked at this morning, which involves several genes whose knockdown disrupted swimming.

[ukemi]

@kltm, this seems to be a GPAD output issue. Maybe we should move this to the Minerva tracker.

[malcolmfisher103]

Maybe that does bring in a commonality since both of these are cases where distinct molecular functions enabled by distinct gene products are linked to the same GO biological process entity.

[ukemi]

That should be allowed. Eventually we want to be able to causally link up all those functions.

[goodb]

@malcolmfisher103 one quick clarification. When you say
" I can still go to the models in Noctua and see the annotations in the 'Workbenches' menu's 'Annotation preview' or 'Export GPAD' option from the 'Model' menu, see this model for an example"

When you use the export to GPAD option or annotation preview right now, are those the GPAD lines that you expect to see? If those are correct, then the difference is some kind of downstream filter in the pipeline. If not, then something is happening with the minerva GPAD generating service. We haven't actively changed Minerva GPAD generation, but lots of other things (e.g. ontology changes) could impact it.

[malcolmfisher103]

Yes, the lines are what I would expect, and what was in the older release version of the GPAD.

[hdrabkin]

Also posted in #335 Alka-Selzer moment
I discovered today that between 6/17 and 6/18, we lost over 50% of our Noctua annotations:
6/17 NOCTUA Annotations:
Total Number of Genes Annotated to: 1027
Total Number of Annotations: 6716

6/18 NOCTUA Annotations:
Total Number of Genes Annotated to: 551 <<<<<<<<<< 476 loss
Total Number of Annotations: 3111 <<<<<<<<<< 3605 loss

[malcolmfisher103]

The new code release from around the 21st seems to have fixed this issue. We have recovered the lost Xenbase Noctua based annotations in the snapshot gpad file.