Abstract

Genome-wide association studies (GWAS) have found widespread evidence of pleiotropy, but characterization of global patterns of pleiotropy remain highly incomplete due to insufficient power of current approaches. We develop fastASSET, an extension of the method ASSET, to allow computationally efficient detection of variant-level pleiotropic association across a large number of traits. We analyze GWAS summary statistics of 116 complex traits of diverse types collected from the NIH GRASP repository and a number of other large GWAS consortia. We identify a total of 2,293 independent loci at the genome-wide significance level and found that the lead variants in nearly all of these loci ([~]99%) to be associated with to two or more (median = 6) traits. Further, the estimated degree of pleiotropy for the detected variants strongly predicted their degree of pleiotropy across a much larger number of traits (K=4,114) in the UK Biobank Study. Follow-up analyses of 21 unique trait-specific variants suggest that they are often linked to the expression in trait-related tissues for a small number of genes, some of which are well known to be involved in relevant biological processes. Our findings provide deeper insight into the nature of complex trait pleiotropy and leads to, for the first time, identification of highly unique trait-specific susceptibility variants.