Load GPI file for UniProt reviewed entries for Virus and Bacteria in NEO

[pgaudet]

Hello,

We'd like to load the UniProt reviewed entries for Virus and Bacteria in NEO.

@alexsign generated a GPI file.
@kltm Where do we make it available?

Thanks, Pascale

[kltm]

@pgaudet Currently, changes to the NEO build are handled over here: https://github.com/geneontology/neo/issues
Once in NEO, we run the build here: https://build.geneontology.org/job/geneontology/job/pipeline/job/issue-35-neo-test/

[pgaudet]

OK - you said go-site on the software call. I'll move the ticket.

Can we assign someone to this ?

[kltm]

@pgaudet Where is this file?

[pgaudet]

In my inbox.

My question was

Where do we make it available?

:)

Thanks, Pascale

[kltm]

Just to follow up on the ticket with conversations help elsewhere:

• we will be provided a URL for a GPI
• we will add this to our metadata
  this is terrible--why do we do this? https://build.berkeleybop.org/job/publish-go-site-datasets-json/
  ideally, we're just getting and working with the YAML product from a pipeline
• we will regenerate the datasets.json used by NEO
  this should self-update, see above
• we will rerun the temporary NEO pipeline
• we will deploy this new NEO

[pgaudet]

This is the URL: ftp://ftp.ebi.ac.uk/pub/contrib/goa/uniprot_reviewed_virus_bacteria.gpi

[pgaudet]

@kltm What do we need to do next ? Can we add this to the GOA yaml ?

[kltm]

@pgaudet That would be the older file at this point. The way the system works current, the Scov2 stuff is bypassing the main system.
We're currently using: https://github.com/geneontology/neo/blob/master/Makefile#L43

[cmungall]

We can go ahead and add this to to neo easily, but there are some things we need to discuss first:

Currently we assume each gpi file is for one species, and we assign a species code. This is for disambiguation of label which is necessary to prevent autocomplete confusion. For example, there are 574 dnaA genes here. I suggest we get around this by appending the taxon id to the symbol. It's not perfect but it's simple to implement.

Second, discussing with @pgaudet we decided that we would come up with a GO reference species list and use this instead. Stay tuned.

Third, we will be loading sars-cov-2 twice (this is configured to be loaded from its own gpi, see geneontology/go-site#1431). I think this is what you are referring to Seth?

[kltm]

@cmungall I'd have to get it all in my head again to be completely clear, but the current use case for the GPI is for annotation in Noctua: in order to make annotation work we need it in NEO; to get it into NEO, we have the somewhat hacked NEO Makefile with the sars-cov-2 handled as a special case. I was responding to @pgaudet 's query about adding an older file to the main pipeline GOA dataset metadata, which would not support this use case.

[kltm]

@pgaudet This is now a dupe on #77 , correct? (Still with the understanding that we may just do #82 anyways.)