So I ran a GWAS on a large sample (n = 480,000) using REGENIE. I wonder does REGENIE uses the lambda GC to correct for genomic inflation, or does it correct for genomic inflation only via some other approach (like specifying PCs as covariates).
I'm also confused about lambda GC in general. It seems like the closer lambda GC is to 1, the less inflated the GWAS is. However, I'm not sure how we can use lambda GC to correct the inflation. In this paper (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3133943/), it is stated that "when lambda is greater than one, all subsequent chi-squared statistics on a set of candidate markers are divided by lambda". My question is how does that help correct the pval?
I know lambda GC is calculated by:
chisq <- qchisq(1-data$pval,1)
lambda <- median(chisq)/qchisq(0.5,1)
say the lambda is 1.6, and the chisq for one of the snp is 3.2. So to do GC correction, the new chisq is 3.2/1.6 = 2. Then do we say the corrected pval for that snp is 0.157? (by solving for pval base on new chisq) Am I correct about this? If yes, then should this process be applied on all pvals, or only on the "set of candidate markers" (which, from my understanding, are the snps with significant pvals)?
Thanks in advance!