Standardize PAINT data production and use in the pipeline

[kltm]

Currently, some resources have their PAINT annotations merged into their main output file (e.g. the paint_pombase.gaf file gets merged into the pombase.gaf file that we provide), while others do not (e.g. there is no paint_japonicus.gaf file, japonicus IBAs are found in the paint_other.gaf file instead).

This easily leads to confusion and makes it hard to diagnose issues.

As a solution, we propose a simple rule: if the primary gene ID space is not uniprot, then always include the paint_xxx file.

I believe this would mean the creation and merging of:

• genedb_lmajor
• genedb_tbrucei
• genedb_pfalciparum
• japonicus
• pseudocap
• sgn
• the filtering of the above from the paint_other.gaf

(I assume that rnacentral is a special case here.)

This stems from the discussion of around a drop in japonicus annotations geneontology/helpdesk#526 .

Tagging @ValWood @cmungall @pgaudet @dustine32

[ValWood]

cc @kimrutherford

[cmungall]

I think we should look at these individually

These are actually the same entity (but it's not obvious!!):

https://amigo.geneontology.org/amigo/gene_product/TriTrypDB:LmjF.22.0870
https://amigo.geneontology.org/amigo/gene_product/UniProtKB:Q4QBP1

We have this dupe problem for all the veupathdbs

However, I do not think we should resolve this by making a new paint file etc

We just take LEIMA_5664_UP000000542.gaf as source of truth. While veupathdb still has a presence, I don't think there are frequent manual annotation updates, and if they need to do this, they should use p2go

[dustine32]

We can start with creating a paint_japonicusdb.gaf file.

I should note the PAINT IBAs are by default output in UniProt unless we add an exception, which we do for the MODs. So, for S. japonicus, we have been using UniProt. I'll change this so it outputs JaponicusDB IDs and I'll need to utilize the JaponicusDB GPI since the UniProt->JaponicusDB mappings were not yet in the Reference Proteome version (2023_05) used to build PANTHER19.0.

[kltm]

@dustine32 Cheers--thank you for the update.

[ValWood]

Will this affect PomBase in any way? Should we now get the PAINT annotations from GO rather than GOA?

CC @kimrutherford

[kltm]

@ValWood This discussion does not affect PomBase at all--only the resources mentioned at the top (with japonicus being a prime motivator).

[ValWood]

Yep, we maintain JaponicusDB

[kltm]

@ValWood With that context, the changes for the GO download products would be as outlined by @dustine32 here: #413 (comment) .

[pgaudet]

@ValWood

Should we now get the PAINT annotations from GO rather than GOA?

I am not clear on the data flow here, but if these are correct when you get them from GOA that should not change.

[ValWood]

I'm not sure. It seems we are not getting them at all, probably because of the incorrect taxon ID. @kimrutherford, could you get the IBA directly from GO? or pull in the strain ID too?

[pgaudet]

This is the file you could load from GOC:

http://snapshot.geneontology.org/products/upstream_and_raw_data/paint_pombase.gaf.gz

Soon a similar file for japonicus will be available.