EPACTS single firth association probllem

[Erkin Acar]

Hello everyone,

I have a problem with firth bias corrected association for single variant analysis. I am running a GWAS with 2150 cases and 2700 controls (with whole genome sequencing data). There have been previous studies on what I am doing, so I am expecting to see similar results. I have split my variants into separate common and rare variant files. If I run the common variants with Wald test (or glm() as mentioned in the documentation), I find the expected results. Just to give an example; one of the very well known variants that is associated with my phenotype has the p-value of 1e-91 in Wald test and effect estimate of ß=-0,98. For the same variant if I run the firth test, my p-value drops to 1e-5 and effect estimate to ß=-0.01. I am okay with not significant results if that's the case but the difference here is huge. I am correcting for gender, principal component 1, and principal component 2 (for the population stratification). Since I see this problem with common variants, I cannot trust the rare variant results in which I cannot really use Wald test as far as I know.

Does anyone have any idea what might be causing this behavior? Just for the record, the previous study that found the variants associated with my phenotype also used EPACTS with firth correction (link here). The example variant I have given has a p-value of 9.6e-618 in this study, with OR 0.38 (that is beta value of -0.97).

Any suggestions and tips are appreciated!

Erkin