RT Journal Article
SR Electronic
T1 Multi-trait genome-wide analyses of the brain imaging phenotypes in UK Biobank
JF bioRxiv
FD Cold Spring Harbor Laboratory
SP 758326
DO 10.1101/758326
A1 Chong Wu
YR 2019
UL http://biorxiv.org/content/early/2019/09/05/758326.abstract
AB Since many genetic variants identified in genome-wide association studies (GWAS) are associated with multiple, sometimes seemingly unrelated traits, one can improve the statistical power of GWAS by testing the association of multiple traits simultaneously. While appealing, most existing methods focus on analyzing a relatively small number of traits and may yield inflated Type I error rates when analyzing a large number of traits. We introduce a new method called aMAT for multi-trait analysis of GWAS summary statistics of an arbitrary number (e.g. hundreds) of traits. We first conduct extensive simulations and demonstrate that aMAT yields well-controlled Type I error rates and achieves robust statistical power when analyzing a large number of traits. Next, we apply aMAT to summary statistics for a group of 58 volume-related imaging phenotypes in UK Biobank. aMAT yields a genomic inflation factor of 1.04 and identifies 28 lead SNPs spanning in 24 distinct risk loci, 13 of which are missed by any individual univariate GWAS. In comparison, the competing methods either yield a suspicious genomic inflation factor or identify much fewer risk loci. Finally, four additional groups of traits have been analyzed and provided similar conclusions.