GAF/GPAD example with chained part_of and causally_upstream_of BP relations - gomodel:586fc17a00000273

[vanaukenk]

Hi,

For discussion on our Wednesday call, I'd like to take another look at the C. elegans copper ion homeostasis model specifically to look at the BP annotations that are returned for the term GO:0009792, embryo development ending in birth or egg hatching, that is linked to GO:0055070, copper ion homeostasis, by a causally_upstream_of_or_within relation.

Right now, the embryo development term is not included in either the AEs or as a separate BP annotation, although I would like it to be.

Thx.

[balhoff]

@vanaukenk should causally_upstream_of_or_within be included as a possible annotation qualifier? Currently this is the list:

• enables
• involved_in
• part_of
• acts_upstream_of
• acts_upstream_of_or_within
• involved_in_regulation_of
• involved_in_negative_regulation_of
• involved_in_positive_regulation_of
• involved_in_or_involved_in_regulation_of

causally_upstream_of_or_within is also not in the extension relations whitelist, but I can add it.

[vanaukenk]

Hi @balhoff

As a qualifier for the GPAD output, the causally_upstream_of_or_within should translate to acts_upstream_of_or_within.

For the annotation extensions list, you're right, I don't see this as an option in Protein2GO. I only see causally_upstream_of and the other usual suspects.

I guess we could go ahead and add causally_upstream_of_or_within to the extensions relations whitelist, but with the caveat that this might need more group-wide discussion.

[vanaukenk]

For the translation between LEGO and GPAD relations, I'm referencing the mapping that @dosumis used in https://github.com/geneontology/go-annotation/issues/1517

[balhoff]

It looks like we have enables o 'causally upstream of or within' -> 'acts upstream of or within'

but I think you would need involved_in o 'causally upstream of or within' -> 'acts upstream of or within'

I can't say offhand whether that would make sense.

[dosumis]

Should be sufficient that part of be a subproperty of 'causally upstream of or within'. I was sure I'd added that in RO, but don't see it in GO.

[dosumis]

For the annotation extensions list, you're right, I don't see this as an option in Protein2GO. I only see causally_upstream_of and the other usual suspects.

I guess we could go ahead and add causally_upstream_of_or_within to the extensions relations whitelist, but with the caveat that this might need more group-wide discussion.

This makes logical sense (e.g. could record that gene product X enables kinase activity that is causally upstream of or within muscle cell differentiation). Definitely needs discussion with annotators though.

[balhoff]

@dosumis making 'part of' a subproperty of 'causally upstream of or within' would make its domain and range both 'occurrent'. I think that would be a problem.

[dosumis]

@dosumis making 'part of' a subproperty of 'causally upstream of or within' would make its domain and range both 'occurrent'. I think that would be a problem.

Of course. Wasn't thinking straight. We've actually been round this before. It's one of a number of cases where we really need to distinguish occurrent part of from continuant part of. That aint gonna happen anytime soon, but we still need the inference. Maybe this could be done using a rule?

[balhoff]

While we can't make part of a subproperty of causally upstream of or within, occurrent part of is already a subproperty of both. I propose we add a SWRL rule to RO:

occurrent(?x) ⋀ 'part of'(?x, ?y) ⋀ occurrent(?y) -> 'occurrent part of'(?x, ?y)

This will "deepen" part of relationships between occurrents to occurrent part of.

Thoughts @dosumis @vanaukenk @ukemi @cmungall?

[balhoff]

Or, we could simply output causally upstream of or within rather than occurrent part of.

[ukemi]

I think the relation chain that we really want here is involved_in o causally_upstream_of -> acts upstream of etc. We'd like to see annotations to the downstream processes. See this model:

http://noctua.geneontology.org/editor/graph/gomodel:5b91dbd100001176

Should we close this ticket and make a new one or just use this one?

[balhoff]

I guess we can keep this ticket open and also open an RO ticket for the desired chains.