Although a large proportion of variability in complex human traits is due to genetic variation, the mechanistic steps between genetic variation and trait are generally not understood1–7. Many genetic variants influence complex traits by modulating gene expression, thus altering the abundance levels of one or multiple proteins8–12. Such relationships between expression and trait could be investigated through association scans in individuals for which both measurements are available8,13,14. Unfortunately, studies that measure gene expression have been held back by specimen availability and cost, with the few published studies of expression and complex trait being orders of magnitude smaller than studies of trait alone. Consequently, many expression-trait associations cannot be detected, especially those with small effects. To mitigate the reduced power from small sample size, alternative approaches examined the overlap of genetic variants that impact gene expression (eQTLs) with trait-associated variants identified in large, independent genome-wide association studies (GWAS)5,6,8,9,11–13,15. However, this approach is also likely to miss expression-trait associations of small effect.
