Include Human Protein complexes in NEO

[pgaudet]

The file is here:

http://ftp.ebi.ac.uk/pub/databases/GO/goa/HUMAN/goa_human_complex.gpa.gz

@kltm please let me know if you need more information.

Thanks, Pascale

[kltm]

Noting that this is a ~6k line compressed GPAD 1.1.

[kltm]

goa_human_complex seems to already be included? https://github.com/geneontology/neo/blob/10210c1e07218f74fa02b86237e672001bffc7de/Makefile#L9 with "source": "ftp://ftp.ebi.ac.uk/pub/databases/GO/goa/HUMAN/goa_human_complex.gpi.gz",

@pgaudet Is there something that you should be seeing that's missing?

[pgaudet]

Thanks, it looks like an autocomplete issue.

If I enter CPX-5993 I dont get anything relevant If I enter CPX-5993 hsap it works.

Is this an otherwise known issue?

@sylvainpoux

Thanks, Pascale

[kltm]

@pgaudet Hm. The entry seems to work as expected on the quick search on http://noctua-amigo.berkeleybop.org (and the entry is clearly there), so it not a NEO issue, but rather a client/autocomplete issue.

I think the issue may be that the label is getting drowned out in the ontology search as it's just a local ID (technically https://github.com/geneontology/amigo/issues/120). Adding the "Hsap" probably boosts up the label enough to get the correct hit. The general search doesn't suffer from this as it also indexes the local ids. Entering the full ID in the ontology search would generally bypass this (e.g. ComplexPortal:CPX-5993).

[sylvainpoux]

Hi,

yes, ComplexPortal complexes can be loaded. Although the autocomplete system is not very convenient, it works. Many thanks

I would have another request.

I made my model (http://noctua.geneontology.org/editor/graph/gomodel:61f34dd300000000). But then realized that it does not pass the Reasoner step. I suspect it is due to the ComplexPortal identifiers.

Do you have any explanation?

Thanks

Sylvain

[vanaukenk]

@kltm How are the goa_human_complex entries from the gpi file above typed in neo and minerva? If they aren't recognized as an instance of GO:0032991 might that explain the ShEx error that @sylvainpoux is seeing?

[kltm]

@vanaukenk @sylvainpoux For questions about the typing (which occurs in minerva, not NEO), I would probably rope in @balhoff .

[balhoff]

There is a NEO connection in that if the GPI containing the terms is not included into NEO, Minerva won't know how to connect them up to the root types.

[vanaukenk]

In the incoming gpi1.2 file, these ComplexPortal entries are typed with the text 'protein_complex'. I don't know how that information gets passed along during the NEO build and subsequent inner workings of minerva.

[balhoff]

I think I see the issue, which is another manifestation of #66 and ultimately #17.

The term in the Noctua model has IRI https://www.ebi.ac.uk/complexportal/complex/CPX-5993

In NEO it has IRI http://purl.obolibrary.org/obo/ComplexPortal_CPX-5993

I'll try adding this namespace similarly to the other fix. @vanaukenk in NEO the term is under information biomacromolecule. Should that work (assuming the identifier issue is fixed)?

[vanaukenk]

@balhoff - technically, these entries should be typed as 'protein-containing complex' (GO:0032991) in NEO.

[balhoff]

@vanaukenk @kltm goa_human_complex.gpi.gz is using object type protein_complex, but the code that maps to GO:0032991 is looking for complex: https://github.com/geneontology/neo/blob/aeb90f98e2517a43fd5440729a1c51944b26497b/gpi2obo.pl#L122

Who is right? Currently GO:0032991 doesn't appear in NEO at all.

[vanaukenk]

Oh, goodness. I'm referring to the ShEx and our GPI2.0 specs.

In ShEx we define ProteinContainingComplex as GO:0032991

[balhoff]

GPI 1.2 spec says it should be Type_Symbol but doesn't define that. Since nothing appears to be matching complex right now, I'm inclined to change the code to match protein_complex.

[vanaukenk]

SGD also includes ComplexPortal ids in their files. I'm double-checking to see what they use for type.

[vanaukenk]

I see protein_complex as well, for them.

[balhoff]

Thanks—I made a PR: https://github.com/geneontology/neo/pull/81

[kltm]

Merged PR; will need to wait to eyeball products once we clear https://github.com/geneontology/neo/issues/80

[kltm]

Continuing build attempts by temporarily shelving #77 and #80

[kltm]

@balhoff @vanaukenk Noting that we've had a NEO cycle with the new code in place.

[kltm]

Talking to @pgaudet , this is closed