Computing Genome Inflation Factor Lambda

Question

[Brian Tsai]

Hi,

I have a list of P-values for SNPs in a genome-wide association study (GWAS), and was trying to figure out how to compute the inflation factor (lambda) that I see reported often in papers. Wikipedia essentially tells me to divide the median test statistic by 0.456, but I only have pvalues here --- is it valid to divide by 0.5, or do I have to do something else, and is there any package (preferably in R) that I can use to compute the inflation?

I've seen the package GenABEL, but for some reason it doesn't just use the median p-value, even when I change a few pvalues at the top (significant) end, the inflation factor changes wildly, so I'm unsure whether the implementation is correct.