Causal models by gene query should look for isoforms

[dustine32]

Sometimes an isoform version of the canonical gene ID is used to annotate a causal pathway model. We should update the GO-CAM API's getGPModelsWith2CausalMFs SPARQL query to look for these and include them in the results for the canonical gene ID.

Ex: SARS-COV2/HOST - displays for human TOMM70 but not for human MAVS. In this model, TOMM70's ID is UniProtKB:O94826 whereas MAVS' ID is UniProtKB:Q7Z434-1. An API call from the Alliance gene page for human MAVS (UniProtKB:Q7Z434) should return causal models for its isoforms (e.g., UniProtKB:Q7Z434-1).

@kltm @sierra-moxon Feel free to move this ticket around given that we're working on migrating the GO-CAM API into the GO API.

[dustine32]

@ukemi Any mouse causal models using isoform IDs?

@deustp01 Noting that this may affect Reactome imports given the eventual shift from REACTO to PRO IDs.

[ukemi]

Not an easy search, but I couldn't find any. If so, they would have entities from PRO. PR:########

[kltm]

I think there is something slightly unsettling about this. I'll mark this for this week's call, but I'm wondering if it might be better to mark the models or GPs with this information explicitly, rather than a "wildcard" search?

[kltm]

@pgaudet I think this is Alliance Support.

[kltm]

TODO:

• check what things are actually loaded in the public SPARQL endpoint
• can this be uploaded into the endpoint
• could this be precomputed?