Genetic analyses of gene expression have great potential to facilitate insights into the genetic basis of complex traits. However, the utility of these data are limited by the extent to which the discovered associations correspond to legitimate, reproducible associations. Our estimates of 49% (UC vs. UW), 57% (UC vs. Merck), and 67% (UC vs. either) cis-eQTL reproducibility are substantially lower than two recent reports between two mouse crosses (76%, [27]), two independent sets of lymphoblastoid cell lines (83%, [25]), and two sets of primary human skin (>99%, [26]). Several non-exclusive possibilities likely contribute to these discrepancies. First, different discovery methodologies and replication criteria were employed in each study. Second, our studies were performed on different expression platforms (Agilent and Illumina), which reduces the influence of reproducible platform-specific errors but may result in missing splice-variant-specific eQTLs [40], [41], [10] as array manufacturers often target different exons in a given gene. However, this is likely to have a limited effect, as we found that the replication rate was not significantly different for genes assessed by probes within the same exon (Figure S10).
