Given the results of our studies on the SNPs reproducibly associated with complex traits, we reasoned that prioritizing follow up of SNPs discovered to show phenotype associations through GWAS should be improved by adding eQTL information to the physical and publicly available functional information commonly used for SNP annotation. Thus, we constructed a score for each SNP that quantifies the likelihood that the SNP has a function (or in strong LD with a functional SNP) in regulating transcript levels. Using the SCAN db file that has annotation information for each SNP (location, host gene, function, CEU MAF) as well as summaries of eQTL p-values (minimum cis p-value, minimum overall p-value), we focused on the 3,844,039 unique autosomal SNPs. The eQTL p-values were truncated at 0.01 for cis-regulatory SNPs (360,696 SNPs with p-value<0.01) and 0.0001 for trans-regulatory SNPs (1,883,759 SNPs). We define a cis-regulatory SNP as a SNP that is within 4 MB of the beginning or end of the gene with which the SNP shows transcript level association, and we define a trans-regulatory SNP as a SNP that is >4 MB from (or on another chromosome than) any gene with which the SNP shows transcript level association.
