Abstract
To understand the biological mechanisms underlying the thousands of genetic variants robustly associated with complex traits, scalable methods that integrate GWAS and functional data generated by large-scale efforts are needed. We derived a mathematical expression to compute PrediXcan results using summary data (S- PrediXcan) and showed its accuracy and robustness to misspecified reference populations. We compared S- PrediXcan with existing methods and combined them into a best practice framework (MetaXcan) that integrates GWAS with QTL studies and reduces LD-confounded associations. We applied this framework to 44 GTEx tissues and 101 phenotypes from GWAS and meta-analysis studies, creating a growing catalog of associations that captures the effects of gene expression variation on human phenotypes. Most of the associations were tissue specific, indicating context specificity of the trait etiology. Colocalized significant associations in unexpected tissues underscore the advantages of an agnostic scanning of multiple contexts to increase the probability of detecting causal regulatory mechanisms.
Prediction models, efficient software implementation, and association results are shared as a resource for the research community.