search

geneontology / minerva

BSD 3-Clause "New" or "Revised" License
6 stars 8 forks source link

causally_upstream_of _or _within -o- part _of -> causally_upstream_of _or _within #527

Closed ukemi closed 3 months ago

ukemi commented 1 year ago

With the decision to eliminate the 'involved in' terms and convert them to GO-CAM models we are losing information when we convert the GO-CAM to standard annotations. If the original annotation had a GP to term relation of causally_upstream_of_or_within 'blah involved in blac' and it is decomposed to an expression in which the GP is causally upstream of or within blah that is part of blac, only a single annotation is generated to blah. The information about the other process is lost in an annotation extension.

We need to generate annotations to both process as happens when the GP2term relation is part_of.

See http://noctua.geneontology.org/editor/graph/gomodel:64d5781900001000 and the annotation preview for an example.

ping @LiNiMGI

balhoff commented 1 year ago

@ukemi are you proposing we add the property chain causally_upstream_of _or _within o part _of -> causally_upstream_of _or _within? I think that makes sense. Could you make an RO issue?

ukemi commented 1 year ago

Yep. https://github.com/oborel/obo-relations/issues/746

ukemi commented 1 year ago

Do you think we also need acts_upstream_of_or_within o part_of -> acts_upstream_of_or_within?

balhoff commented 1 year ago

Do you think we also need acts_upstream_of_or_within o part_of -> acts_upstream_of_or_within?

Probably so.

kltm commented 1 year ago

@ukemi Is this part of a project, or something else?

ukemi commented 1 year ago

Something else. Unless we consider annotation review due to obsoletion a project.

kltm commented 1 year ago

@ukemi Nothing to worry about at your end as it's just metadata, but anything that touches code or the data store (which I believe this is?) should probably be grouped somehow. We have other "projects" for reviews, so something could be made here. I can touch bases w/ @pgaudet when she's back.

ukemi commented 1 year ago

@balhoff knows better than I but I suspect once the property chains are added to RO, the reasoning that is in place will consider them and create the missing annotations.

balhoff commented 1 year ago

@balhoff knows better than I but I suspect once the property chains are added to RO, the reasoning that is in place will consider them and create the missing annotations.

That's true, it should only require changes in RO.

pgaudet commented 1 year ago

If this only requires changes in RO, should we close this ticket?

balhoff commented 1 year ago

I opened a pull request: https://github.com/oborel/obo-relations/pull/771

balhoff commented 3 months ago

This has been merged in RO.