Assembling KG-PHENIO...from scratch!

[caufieldjh]

Well, not entirely from scratch. Mostly from KG-OBO.

[caufieldjh]

One difference between assembling PHENIO in KGX is handling of the bridge and equivalency files - these are essentially mappings of one class to another. They could be ingested as xrefs here but:

• KGX doesn't know how to do that, though having them in SSSOM may help
• would it make any difference in the graph? We don't want the xrefs to exist as distinct edges, but we do want to recognize equivalency across namespaces.

The NCBI taxonomy slim used by PHENIO also isn't on KG-OBO, so I'm just using the same file here (http://purl.obolibrary.org/obo/ncbitaxon/subsets/taxslim.owl)

[caufieldjh]

Trying to use upheno2 in PHENIO, so that should correspond here

[caufieldjh]

This unfortunately won't quite work without the upstream processing (i.e., KG-OBO) incorporating some of the changes from #43, namely expanding relations based on subq patterns

[matentzn]

One difference between assembling PHENIO in KGX is handling of the bridge and equivalency files - these are essentially mappings of one class to another. They could be ingested as xrefs here but:

Good point - Maybe the way forward here is to have Uberon, uPheno and Mondo provide the SSSOM mappings directly to replace the "bridge file" mechanism for this use case.

Maybe a hackathon between a handful of people is in order, as this seems like a general issue we will be facing a lot