feat(study_locus): remove statistics after conditioning from schema

[ireneisdoomed]

This PR includes:

• deletion of the variant statistics after conditioning from the StudyLocus schema

Context

Inside a locus, we have a couple of statistics that describe the association between a variant and the trait after doing conditional analysis.
COJO does conditioning, and that is why we were maintaining these fields. However our current implementation does not include adjusting the p values based on other signals in the locus.

Given the current size of our locus object, I think we're better off without these null fields polluting the codebase.

[addramir]

I agree with the changes. But if we want to get rid of COJO stats we need to also rename - "logABF" to "logBF" in study_locus.json

[ireneisdoomed]

if we want to get rid of COJO stats we need to also rename - "logABF" to "logBF"

@addramir Thank you for the comment, one question about colocalisation. The COLOC implementation i know uses ABFs to do the hypothesis testing (see here). Can I double check that we can just exchange ABFs with BFs and achieve the same?

[addramir]

Yes, you can. It is basically the same, just some terminological issues.

[addramir]

As discussed with Irene, for changing ABF to BF it is better to open another PR.