Create a GO-CAM -> TSV file for end user consumption

[kltm]

Create a GO-CAM -> TSV file for end user consumption. This could fill the ecological niche of our previous SIF effort.

TBD:

• [ ] is this a one-off, or is this a future pipeline product?

Tagging @dustine32 @balhoff

[kltm]

Also tagging @pgaudet

[kltm]

Noting for @balhoff, technically, not having the SPARQL endpoint or store involved (i.e. a file-to-file transformation) would likely be easier to reuse and recreate. (Dealing with the endpoint complicates things and the blazegraph is--or should be--a moving target right now.)

[balhoff]

@cmungall what are the desired columns? Something like this (using noted property path)?

• subject
• subject_activity (?mf enabled_by s)
• subject_location (?cc ^occurs_in/enabled_by s)
• subject_process (?bp ^part_of/enabled_by s)
• relation (?s ^enabled_by/?rel/enabled_by ?o)
• object
• object_activity
• object_location
• object_process
• model_iri

Taking this model as an example: http://noctua.geneontology.org/editor/graph/gomodel:645d887900000758?

For hormone activity enabled by BGLAP, would we only include the brain development process, and not the other two which it's part of?

[balhoff]

Sample table for http://noctua.geneontology.org/editor/graph/gomodel:5ee8120100000524:

?gp1	?mf1	?cc1	?bp1	?relation	?gp2	?mf2	?cc2	?bp2	?model
http://identifiers.org/uniprot/O15049	http://purl.obolibrary.org/obo/GO_0005515			http://purl.obolibrary.org/obo/RO_0002629	http://identifiers.org/uniprot/P46934	http://purl.obolibrary.org/obo/GO_0061630			http://model.geneontology.org/5ee8120100000524
http://identifiers.org/uniprot/Q7Z434	http://purl.obolibrary.org/obo/GO_0035591			http://purl.obolibrary.org/obo/RO_0002629	http://identifiers.org/uniprot/Q12933	http://purl.obolibrary.org/obo/GO_0031625			http://model.geneontology.org/5ee8120100000524
http://identifiers.org/uniprot/P46934	http://purl.obolibrary.org/obo/GO_0061630			http://purl.obolibrary.org/obo/RO_0002629	http://identifiers.org/uniprot/Q7Z434	http://purl.obolibrary.org/obo/GO_0035591			http://model.geneontology.org/5ee8120100000524

[balhoff]

And for the BGLAP model mentioned above:

?gp1	?mf1	?cc1	?bp1	?relation	?gp2	?mf2	?cc2	?bp2	?model
http://identifiers.org/uniprot/P02818	http://purl.obolibrary.org/obo/GO_0005179	http://purl.obolibrary.org/obo/GO_0005576	http://purl.obolibrary.org/obo/GO_0007420	http://purl.obolibrary.org/obo/RO_0002413	http://identifiers.org/uniprot/Q5T848	http://purl.obolibrary.org/obo/GO_0004888	http://purl.obolibrary.org/obo/GO_0005886	http://purl.obolibrary.org/obo/GO_0007420	http://model.geneontology.org/645d887900000758
http://identifiers.org/uniprot/P38435	http://purl.obolibrary.org/obo/GO_0008488		http://purl.obolibrary.org/obo/GO_0017187	http://purl.obolibrary.org/obo/RO_0002630	http://identifiers.org/uniprot/P02818	http://purl.obolibrary.org/obo/GO_0005179	http://purl.obolibrary.org/obo/GO_0005576	http://purl.obolibrary.org/obo/GO_0007420	http://model.geneontology.org/645d887900000758

[kltm]

@balhoff Curious: would end users be wanting URIs or CURIEs? Guess it depends on the audience?

[balhoff]

@kltm just an information example for now. We can do curies in the end. But I want to check that I'm pulling out the right stuff.

[cmungall]

For hormone activity enabled by BGLAP, would we only include the brain development process, and not the other two which it's part of?

I think having a pipe separate list for this should be fine (it should always be a tree structure and hence the list can always be interpreted as a chain)

It may turn out that this is overkill and there is not so much information in nested part-ofs. Looking at that model, one of the paths is clearly wrong (unless neurotransmitters can think). But as a first pass, having this be transparent is a great way for us to easily spot check some of these

[balhoff]

@cmungall specifying the order of transitive part-ofs might require a different approach; I'm using SPARQL right now, nice and simple but property paths don't capture that. If we're okay with interpreting as a bag of relevant terms then we can use the property paths.

[balhoff]

With some BP and CC aggregation (unordered):

?gp1	?mf1	?cc1s	?bp1s	?relation	?gp2	?mf2	?cc2s	?bp2s	?model
UniProtKB:P02818	GO:0005179	GO:0005576	GO:0007420,GO:0050890,GO:0001956	RO:0002413	UniProtKB:Q5T848	GO:0004888	GO:0005886	GO:0007420,GO:0050890,GO:0001956	gomodel:645d887900000758
UniProtKB:P38435	GO:0008488		GO:0017187	RO:0002630	UniProtKB:P02818	GO:0005179	GO:0005576	GO:0007420,GO:0050890,GO:0001956	gomodel:645d887900000758