Eqtl Results Adjusted For Genomic Inflation Factor

Question

[Zhenyu Zhang]

I've used MatrixEQTL to find eQTLs and got the p-values. I ran these p-values through GenABEL to get inflation factor lambda. However, what to do next? Based on GWAS analysis, I should divide each Chi-square statistics with this factor, regenerate p-values, and calculate q-values, is it correct? However, for eQTLs, I only have t-stat, should I use t-stat to do the same thing (new_pval = pt ( -abs ( t-stat / lambda ), df ) * 2 ) ? Thanks a lot.

Answer 1

[Fabio Marroni]

I think that the most correct thing is the following, I hope that someone can confirm (or contradict) this:

1) Transform all your pvalues in chi-square values with one df using qchisq function, and set lower.tail=FALSE

2) Run genabel to estimate lambda (you can estimate lambda based on chisquare values as well)

3) Divide the chisquare values by lambda

4) Compute the pvalues from the new chisquare values using the function pchisq, again setting lower.tail=F