search

geneontology / go-ontology

Source ontology files for the Gene Ontology
http://geneontology.org/page/download-ontology
Creative Commons Attribution 4.0 International
223 stars 40 forks source link

Clarify primary source for GOREL axioms #14827

Open balhoff opened 6 years ago

balhoff commented 6 years ago

I've seen a few, possibly redundant, sources for axioms about GOREL properties. There are some included directly in go-edit.obo, e.g.:

[Typedef]
id: mediated_by
name: mediated_by
namespace: external
xref: GOREL:0001007
is_a: has_participant ! has participant

There are some GOREL axioms in imports/ro_pending.owl. This file is included in the GO owl:imports. There are also some axioms about terms in the RO namespace in this file, and there doesn't seem to be any means to update these relative to the current version of RO.

There are some GOREL axioms in extensions/gorel.owl. However, it doesn't look like this file is ever imported into GO. There is a gorel-edit.owl, but the Makefile doesn't seem to contain any targets which use this to produce gorel.owl.

balhoff commented 6 years ago

There is a gorel-edit.owl, but the Makefile doesn't seem to contain any targets which use this to produce gorel.owl

Oh, I see this part is fixed by #14700.

balhoff commented 6 years ago

The main question I am after is whether the GOREL axioms can be removed from go-edit.obo, so that these cannot get out of sync. But if they're needed in the ontology, then gorel.owl probably should be imported.

ukemi commented 6 years ago

Hi @balhoff . In an ideal world, there wouldn't be separate relations for the GO annotation extensions and everything would come from a single source (RO). How many and what relations would get lost if we removed them from go-edit.obo? Shall we have a look at these in detail on Monday's call?

Happy New Year!

balhoff commented 6 years ago

Happy New Year! I don't intend to lose any from go-edit.obo, but instead get them via an import. I've done this in #14825 for all the RO relations.

They could even have their primary home in go-edit.obo, such that gorel.obo is produced as an extract (just an idea). But currently, I don't think edits to gorel-edit can automatically propagate to go-edit. And I'm a little confused about the relationship between gorel and ro_pending, which is actually imported by go-edit.

This is all related to being able to produce a file which contains all the GO axioms, no external axioms, and no inferences, to avoid cyclic dependency problems when using reasoners, as in https://github.com/ontodev/robot/issues/38.

ukemi commented 6 years ago

Can you ever see a case where the imports from gorel and ro would clash or be inconsistent? It seems dangerous to me to be getting relations from two different sources. Doesn't it make more sense to have gorel have their home in ro instead of on their own or in go-edit.obo? Maybe I am missing something.

pgaudet commented 5 years ago

@balhoff Is there a resolution to this ? Should we not switch to RO whenever possible ?

balhoff commented 5 years ago

There are still several GOREL relations used in go-edit.obo. Some of this should be cleaned up by #18085. I think we should have a goal to switch to RO and stop using GORELs in go-edit.obo. Should I just retitle this issue with that goal in mind?