"contributes to" in sgd.gaf

[murphyte]

Hi GOA/SGD,

The sgd.gaf download file:
http://current.geneontology.org/annotations/sgd.gaf.gz

Recently added the phrase "contributes to" in column 4, in addition to some other rows using "contributes_to". Should this be standardized, and if so, which way?

Similar for "colocalizes with" vs "colocalizes_with".

Thanks for checking!

-Terence

[srengel]

The rows from SGD are 'contributes to'. The gaf provided by SGD is consistent on all qualifiers (no underscores). The rows added by GOC that are source=GO_Central seem to be the offenders (using the underscores).

Yes, it should be standardized. The rows added by GOC that are source=GO_Central need to be corrected to use the correct format (no underscore).

[murphyte]

I'll note that historically the file only contained "contributes_to", including rows from SGD. I didn't dig through revisions, but I think it's a recent change given that our code only just now failed because of the change.

I think all the other files still use "contributes_to", but I didn't check if those are also all sourced from GO_Central, or if there are other sources also using the underscore.

[suzialeksander]

commenting to keep an eye on this

[kltm]

@murphyte @srengel @pgaudet @dougli1sqrd
I've been doing a little digging and I wanted to clarify a few things. From the most recent release:

zgrep -c "contributes to" sgd-src.gaf.gz
1278
zgrep -c "contributes_to" sgd-src.gaf.gz
0
zgrep -c "contributes to" sgd.gaf.gz
1037
zgrep -c "contributes_to" sgd.gaf.gz
90
zgrep "contributes_to" sgd.gaf.gz | grep -c IBA 
90

What this means is that all "contributes to" for this release are coming from the SGD source, with the "contributes_to" in the post-GO pipeline product being added from PANTHER.
Apparently owltools is upgrading "contributes to" to "contributes_to", so this issue is not seen in AmiGO.

In December:

zgrep -c "contributes to" sgd-src.gaf.gz
0
zgrep -c "contributes_to" sgd-src.gaf.gz
1961

So it seems that SGD used to use "contributes_to", then switched to "contributes to" within the last few months. According to the GAF spec (http://geneontology.org/docs/go-annotation-file-gaf-format-2.1/#qualifier-column-4), the qualifier field is: "one (or more) of NOT, contributes_to, colocalizes_with".

[pgaudet]

Yes ! This should have been picked up in the rules.
I think I checked 'contributes' :(
gorule-0000001 is supposed to be checking this, see
http://release.geneontology.org/2020-03-25/reports/ecocyc-report.html#gorule-0000001

I think some common errors (case, underscore) should be fixed automatically.

@kltm can we make the qualifier its own separate rule ?

Thanks, Pascale

[kltm]

@pgaudet If it would be useful as a rare and separate catch--I think these recent developments are quite exceptional--adding a separate rule to be worked out with @dougli1sqrd would be fine. Generally, it would be simply an obviously illegal value that should fail the most basic parsing (0000001). This was nicely demonstrated here #252 (comment)

[suzialeksander]

Methinks the new releases have fixed this specific issue, thanks @murphyte for catching it. Closing, as this ticket seems to have spawned some new tickets and the specific issue is fixed.