search

geneontology / gp2term

1 stars 0 forks source link

Current results (IEA + IBA) #1

Open dustine32 opened 6 years ago

dustine32 commented 6 years ago

results_2018-08-28.tar.gz Finally got some recent results to show (attached). This should contain matches from IEA and IBA BP annotations.

Still need to start attacking the "IMP, IDA, IGI, IPI, IEP, ISS (and children)" annotations. @vanaukenk @pgaudet @ukemi Let me know what I got wrong so far!

vanaukenk commented 6 years ago

@dustine32

I've been looking through the WB annotations and just wanted to double-check about a gene that I would expect to be in the IEA + IBA files, but can't find it when searching. Maybe I am missing it, but the gene is slo-1.

I see InterPro2GO-based annotations in the latest WB snapshot GAF release, for example:

WB WBGene00004830 slo-1 GO:0005216 GO_REF:0000002 IEA InterPro:IPR005821 F Y51A2D.19|nsy-3 gene taxon:6239 20180416 WB

WB WBGene00004830 slo-1 GO:0006811 GO_REF:0000002 IEA InterPro:IPR005821 P Y51A2D.19|nsy-3 gene taxon:6239 20180416 WB

Can you take a look just to make sure we're not missing things we should have? Thx!

dustine32 commented 6 years ago

@vanaukenk Awesome find! Yeah, these should be getting paired up but aren't with the code that I have going right now. Although, when using just these two lines as input, they do get paired together. Weird. I'll keep debugging, thanks!

dustine32 commented 6 years ago

@vanaukenk OK, I figured it out (this example helped find three bugs!). I'll regenerate the file set and attach them below. Let me know if I should really create a new issue for each updated result set.

dustine32 commented 6 years ago

results_2018-09-10.tar.gz Using GO release GAFs from 9/6/18.

vanaukenk commented 6 years ago

Thanks @dustine32 I'll take a look at the newly regenerated file set and let you know how things look. I think it's okay for now to keep this all in one ticket.

vanaukenk commented 6 years ago

@dustine32 - thanks again for the output files. Looking over the outputs, I think we should now go ahead with the next part of the heuristics, i.e. applying the 'involved in' relation to any experimentally generated annotations that either exactly match or are a child of an InterPro2GO IEA or a PAINT IBA annotation.
I caught a few outliers, mainly for the InterPro2GO annotations, but I think they were unlikely to affect the next round. For this next round we can use both the InterPro2GO IEA and PAINT IBA annotations in the matches file as well as the leftovers file and follow the same output file format of matches and leftovers. Let me know if you want to touch base at all about this next round. Thx!