Decide on format for find and replace tsv

[vanaukenk]

This ticket is to define the fields we want in a TSV that will be used to apply find and replaced_by changes in GO-CAMs outside of the changes that can be made based on replaced_by tags in an ontology. For example, this TSV would handle changes to RO terms used based on changes in annotation practice.

Here are 4-5 fields I can think of to include; there may be others:

1. Existing term
2. Replacement term
3. Github handle of requester
4. Link to relevant github ticket
5. Date requested (maybe that could somehow be included automatically?)

@kltm @balhoff @cmungall - anything I'm missing or that you think is not needed? Thx.

[balhoff]

@vanaukenk I would like to have two TSVs: replace_relations.tsv and replace_classes.tsv. The database change code is pretty different for the two cases. I think that format sounds sufficient.

[balhoff]

My preference is to specify terms in the TSV by IRI and not by CURIE, e.g. http://purl.obolibrary.org/obo/GO_0036268 and not GO:0036268. This removes a dependency on which prefixes are configured in Minerva. Does anyone object to this?

[kltm]

@balhoff I suppose I'd ask how difficult/annoying that would be?
Generally, with all of the formats we consume and interact with, we use CURIEs. I think it would be nice to continue that here, unless there is an overriding difficulty or other concern.

(I'd hope that the cases we're dealing with are narrow enough we wouldn't need to involve larger mapping mechanisms.)

[balhoff]

That's fine, it won't be hard to handle. I suppose the CURIE ship has sailed. :-)

[vanaukenk]

Thanks @balhoff and @kltm
So, we will go with CURIEs and have two separate tsv files, one for relations and the other for classes.

[balhoff]

I've pretty much got this command implemented now. One thing to decide is whether to use CURIEs in comments instead of full IRIs like I did for replace-obsolete. I thought of it as being more unambiguous, but I don't think others like it, so now I have a way to put CURIEs in the comments. If we go with CURIEs here, I think we should update the replace-obsolete command and also run a SPARQL update to edit the existing comments.

[kltm]

@balhoff I think that might be nice. Maybe we could touch bases at some point with @vanaukenk to talk a little about this and update-related (geneontology/noctua#766).

[vanaukenk]

@kltm @balhoff
Wrt scheduling the relations updates, let's plan for the Noctua systems outage on May 12th. That will give us time to do a test run on noctua-dev to make sure everything's working as expected.

[balhoff]

@kltm here are running instructions: https://github.com/geneontology/minerva/pull/478/files#diff-aefb43779e951b6f961416f1195b5d9b753eca8a12a29f64a7a2cd8fb432e63d

[vanaukenk]

This work is done.