d0choa
commented
2 days ago A revised decision:
- PICS will filter for 99%
- SuSie will produce 99% credible
Validation step:
- it will annotate the credible sets
- filter the credible sets for 99%
Subsequent steps:
- they will not filter the credible sets further
We will not change the schemas now for convenience, but i might be a good thing to revise.
Warning: the current PICS results are already filtered by 95% due to a filter that does not seem to be in the code. Hopefully, this threshold won't be there next time we run PICS
As discussed with @addramir and @DSuveges, we would like to be conservative in the release process and implement a 95% credible set filter during Study Locus validation.This will imply that a more strict definition of the credible set will be applied to downstream processes (e.g. Colocalisation), and we will only display 95% (more stringent) credible sets in the FE application (cc @gjmcn).The schemas will remain the same across gentropy + API, and the 99% credible sets will still be available in the staging buckets for future benchmarking exercises.