I'm trying to carry out eQTL analysis using genotype calls from SNP6 array(calls made using CRLMM) and expression from RNAseq from TCGA. I carried out routine QC and population stratification on the genotype data. I also filter gene expression data to remove genes that are with 90th percentile expression < 30 and transform the expression data to a normal distribution while retaining rank as advised in the tutorial for MatrixEQTL. I have a huge inflation in p-values (very very low p-values in the order of e-300) when I carry out eQTL analysis. I have tried adding upto 150 covariates using PEER, however it does little impact on the p-values I obtain. I also tried correcting the expression data for copy number effects by using residuals obtained after regression expression against CN as descried in http://www.cell.com/cell/fulltext/S0092-8674(12)01556-5.
I will be grateful for any advice on how I can correct my expression or genotype data to improve eQTL calling.
Thanking You, Vakul