Collapsing regulates relations

[dustine32]

From 7/23/18 email with Paul:

Hey Paul!

So my question about the Signor models comes from when we looked at one of the PomBase models (for BP GO:0006351 - transcription, DNA-templated). I recall you mentioning that we can collapse/simplify regulates relations like these:

But that we should only do this simplification in the Signor models since we have control over how to infer their data into models. This above set of regulates relations looks like it can be sensibly reduced since both terms are likely related in the ontology and they're connected to the same nodes.

But for Signor there always seems to be some difference that makes me hesitate to combine relations:

Starting with HHIP towards the bottom right, it enables two protein binding activities, both of which are positively regulated by the MF root enabled_by GLI2 in addition to the MF root enabled_by GLI1. So these HHIP protein binding nodes seem to have the same interactions, however, one directly negatively regulates SHH and the other directly negatively regulates DHH. Can we still reduce these two protein binding nodes to just one node that directly negatively regulates both SHH and DHH? Or can each activity only have one outgoing regulates edge?

Let me know if you want me to clarify anything further. Thanks!

-Dustin

Paul's response:

For now, let’s try something really simple: collapse multiple MF nodes into one “union” node if they are enabled by the same gp, and are exactly the same GO MF term. The union node will of course be that same GO term. The incoming edges to the new “union” node will be the set of all incoming edges to the set of collapsed nodes, and ditto for the outgoing edges. That way the pombase ones will not collapse since they are different go terms, but the signor ones will.