Determining the biological significance of findings from genome-wide association studies (GWAS) has emerged as a major challenge for complex trait analysis, as over 90% of significant associations are non-coding. Several lines of evidence suggest that genetic variation implicated in GWAS alters transcription1–3. Expression quantitative trait loci (eQTLs)4,5 overlap markedly with GWAS-identified SNPs, both collectively6–8 and for specific traits (e.g., height, adiposity, cardiovascular risk factors, chemotherapy-induced cytotoxicity, autism, schizophrenia, and Crohn’s disease).9–16 An estimated 55% of eQTL SNPs lie in DNase I hypersensitivity sites and 77% of significant GWAS SNPs are in or correlated with these sites.2,17,18 Although understanding of eQTLs has progressed rapidly, important questions remain. Most eQTL catalogs are incomplete and few studies have had sample sizes n>1000,15,19,20 while n > 3,000 may be necessary for more complete eQTL identification.21 Many eQTLs do not replicate, even using the same HapMap lymphoblastoid cell lines (LCLs) under standardized procedures.19 Replication of distant (trans) eQTLs has been particularly elusive.22 Potential sources of variation including tissue,8,10,23 ancestry,7 winner’s curse effects, and batch effects,5,7,24–26 and cell heterogeneity.27,28
