search

rgcgithub / regenie

regenie is a C++ program for whole genome regression modelling of large genome-wide association studies.
https://rgcgithub.github.io/regenie
Other
187 stars 55 forks source link

Different P-values between REGENIE and PLINK software #445

Open FMinnai opened 1 year ago

FMinnai commented 1 year ago

Hi everyone,

I performed an analysis with REGENIE on 2058 patients with a quantitative trait and 7 different covariates. For 5 SNPs I reached a p-value < 5x10-8.

I repeated the linear regression with PLINK and these 5 SNPs are always the top significant, but with a p-value around 1x10-6.

To perform a check, I used another linear phenotype and I performed the GWASes. P-values from both the two softwares are quite similar (around 1x10-6).

I was wondering the reason of this difference in p-values, if there is any trouble maybe (probably related to the genetic predictors in the Step 1?), and which software I can trust. Thank you for your kind help and suggestions,

Best regards, Francesca

joellembatchou commented 1 year ago

Hi Francesca,

Could you send the logs for the REGENIE and PLINK runs?

Cheers, Joelle