Our analysis was initially confined to probes that had a midpoint distance to the tested SNPs less than 500 kb. Analysis of the significant cis-eQTLs SNP-probe distances (Figure 1) suggests few cis-eQTLs have been missed by imposing this threshold, as in both datasets for 95% of the cis-eQTLs, the SNPs map within 250 kb of the probes. As such it is expected that an increase in statistical power can be achieved by reducing the distance to 250 kb as less tests will be performed. Indeed in the celiac peripheral blood dataset more cis-eQTLs were identified (in total 2,487 different SNP-probe effects were observed, compared to 2,178 in the 500 kb analysis, based on an FDR of 0.05), reflecting 765 unique probes and 2,315 unique SNPs. For the HapMap B cell line dataset 3,226 different SNP-probe effects were identified (compared to 2,839 in the 500 kb analysis, based on an FDR of 0.05), reflecting 994 unique probes and 2,826 unique SNPs (see Table 1). Subsequent analyses were confined to cis-eQTLs that had been detected using this 250 kb window analysis.
