define workflow for feeding new HP-MP link predictions to Nico et al for manual curation/inspection, and incorporation back into Phenio

[justaddcoffee]

Per convo with @matentzn, we should define a workflow for feeding new HP-MP link predictions (i.e. HP-MP link that do not currently exist in kg-phenio) so that they can be evaluated by Nico and manual curators.

@caufieldjh

[justaddcoffee]

Possible workflow:

• write NEAT.yaml to train and apply classifier(s) for MP-HP link prediction to produce TSV (or SSSOM) file with link that don't exist in KG-PHENIO
• ping Nico et al with URL of this file
  -write file (in KGCL possibly?) for suggested incorporation of new MP-HP edges that seem good to curators
• incorporate new MP-HP edges into Phenio uPheno, eventually to be ingested by KG-PHENIO

@cmungall thoughts?

[caufieldjh]

See also Knowledge-Graph-Hub/neat-ml#90 - producing SSSOM and storing it in a consistent location takes care of the first two of the above tasks

[matentzn]

write file (in KGCL possibly?) for suggested incorporation of new MP-HP edges that seem good to curators

Dont worry about this, this will be taking care of on the uPheno side!

[matentzn]

incorporate new MP-HP edges into Phenio, to be ingested by KG-PHENIO

So is this, uPheno, not KG-PHENIO

[justaddcoffee]

Okay, thanks. I think ultimately we'd like to incorporate this back into KG-PHENIO - sounds like new edges would go into uPheno, which would then go into Phenio, which would then go into KG-PHENIO (right?)

[matentzn]

Exactly! :)