As the eQTLs we identified are associations between effectively anonymous SNPs and expression of a nearby gene, we were also interested in fine-mapping the associations, ideally to a causal variant (expression quantitative-trait-nucleotide or eQTN) or haplotype. We therefore re-sequenced the promoter and 3′UTR sequences for 18 genes with strong cis-eQTLs within the 60 UW livers (Table S3). Thirteen of these genes harbored a common SNP or indel within the proximal promoter or 3′UTR that correlated strongly (p-value<1×10−8) with the expression level of that gene, while 17 of 18 harbored a variant with at least a modest correlation (p-value<0.001). Of these 17 genes, the most strongly correlated SNP was within the 3′UTR for 11 genes and within the promoter region for 6 genes. Moreover, 10 of the 17 best SNPs were not within HapMap, indicating that a majority of the most strongly associated promoter/3′UTR variants were neither genotyped directly nor imputed and therefore not detectable in the original eQTL analysis.
