Newly seen in gpad files, qualifier, "NOT|", should be "NOT"

[dvklopfenstein]

Thank you for the great gene ontology annotations.

I am seeing many qualifiers which look like this:

NOT|

rather than what they were in the past:

NOT

This is causing our gpad reader to fail. I am going to change the gpad reader to not fail upon encountering an unnecessary divider, but the large annotation files already take 15-30 seconds to load on my machine and adding this extra check will extend that time.

It would be cleaner if when there is only one qualifier to not include the divider.

I am seeing this in the goa_human.gpad downloaded on March 17, 2020.

[pgaudet]

@dvklopfenstein Sorry about the late reply - where did you download the files from ?

Thanks, Pascale

[dvklopfenstein]

Thanks for looking into this. I should have added the download source when creating this issue. I still have the files from March 17 on my machine:

$ l goa_human.g*
-rw-r--r--+ 1 note2 note2 84323582 Mar 17 13:49 goa_human.gaf
-rw-r--r--+ 1 note2 note2 55388353 Mar 17 13:50 goa_human.gpad
-rw-r--r--+ 1 note2 note2 63116193 Mar 17 15:47 goa_human.gpa

It is interesting to see the goa_human.gpad is different from the goa_human.gpa file.

The goa_human.gpa files work fine and come from ftp://ftp.ebi.ac.uk/pub/databases/GO/goa

The goa_human.gpad file, downloaded from http://current.geneontology.org/annotations
has the trailing divider, NOT|, which caused us to change our software so we could read the file successfully.

[pgaudet]

@kltm Can we fix this ASAP ?

[kltm]

Talking to @dougli1sqrd , fix in progress.

[kltm]

@pgaudet We may want to add this to the release notes for the last release.

[kltm]

@dougli1sqrd Noting points of interest in:
/ontobio/model/association.py
/ontobio/io/assocwriter.py

[dvklopfenstein]

Just wanted to make sure that you noticed that the content of goa_human.gpa and goa_human.gpad are different. For example:

$ grep NOT goa_human.gpad | sort | grep Q9ULI4
UniProtKB       Q9ULI4  NOT|    GO:0003777      GO_REF:0000024  ECO:0000250     UniProtKB:Q52KG5                20091119        UniProt
UniProtKB       Q9ULI4  NOT|    GO:0007018      GO_REF:0000024  ECO:0000250     UniProtKB:Q52KG5                20091119        UniProt
UniProtKB       Q9ULI4  NOT|    GO:0007018      PMID:21873635   ECO:0000318     PANTHER:PTN000648413|PANTHER:PTN000649056               20170228        GO_Central
UniProtKB       Q9ULI4  NOT|    GO:0016887      GO_REF:0000024  ECO:0000250     UniProtKB:Q52KG5                20091119        UniProt
UniProtKB       Q9ULI4  NOT|    GO:0016887      PMID:21873635   ECO:0000318     PANTHER:PTN000648413|PANTHER:PTN000649056               20170228        GO_Central

$ grep NOT goa_human.gpa | sort | grep Q9ULI4
UniProtKB       Q9ULI4  NOT|enables     GO:0003777      GO_REF:0000024  ECO:0000250     UniProtKB:Q52KG5                20091119        UniProt         go_evidence=ISS
UniProtKB       Q9ULI4  NOT|enables     GO:0016887      GO_REF:0000024  ECO:0000250     UniProtKB:Q52KG5                20091119        UniProt         go_evidence=ISS
UniProtKB       Q9ULI4  NOT|enables     GO:0016887      PMID:21873635   ECO:0000318     PANTHER:PTN000648413|PANTHER:PTN000649056               20170228        GO_Central              go_evidence=IBA
UniProtKB       Q9ULI4  NOT|involved_in GO:0007018      GO_REF:0000024  ECO:0000250     UniProtKB:Q52KG5                20091119        UniProt         go_evidence=ISS
UniProtKB       Q9ULI4  NOT|involved_in GO:0007018      PMID:21873635   ECO:0000318     PANTHER:PTN000648413|PANTHER:PTN000649056               20170228        GO_Central              go_evidence=IBA

And the count of NOTs seen in the files is different too:

$ grep NOT goa_human.gpad | wc -l
1244

$ grep NOT goa_human.gpa | wc -l
1250

[kltm]

@dvklopfenstein Fixes in the snapshot should be out tonight on the success of the pipeline. If so, we'll proceed to attempting a new release.

[kltm]

This appears to now be fixed in the current snapshot.