search

opentargets / issues

Issue tracker for Open Targets Platform and Open Targets Genetics Portal
https://platform.opentargets.org https://genetics.opentargets.org
Apache License 2.0
12 stars 2 forks source link

Refactor the logic of training the model #3525

Open addramir opened 2 months ago

addramir commented 2 months ago

We still didn't decide the way how we are going to train the model. Is it going to be the old way? In this case how we select studies? On on EFO level? In this case what to do with features like number of snps in credible set?

addramir commented 2 months ago

Image This is the latest agreed logic of how we should train the model. The current paradigma is "Best credible set per EFO". Credible set level FM -> Best credible set per EFO FM