We thus tested the null hypothesis that SNPs classified as eQTLs for the genes ranked at the top of the list based on the evidence for differences in expression levels between individuals with high or low lymphocyte count are not more likely to be ranked at the top of the independent list based on the strength of the genetic association with lymphocyte count. Because slightly different sets of SNPs met quality control thresholds in the two studies, we limited the analysis to 10 239 of the 11 282 expressed genes (see Materials and Methods for details). To determine the number of gene-SNP pairs to be considered in this analysis, we calculated the median GWAS P-value for increasingly larger subsets of top-ranked genes. Using this approach, we were able to reject the null hypothesis. For example, we found that for the top 33 genes, the median GWAS P-value of the associated cis eQTL SNPs is 0.28, which is much lower than the genome-wide (i.e. for all cis eQTL SNPs) median GWAS P-value of 0.50 and significantly lower than expected for a
