Abstract

Genome-wide association studies (GWAS) have successfully identified tens of thousands of genetic variants associated with various pheno-types, but together they explain only a fraction of heritability, suggesting many variants have yet to be discovered. This is because the genetic effects are typically very modest, and a much larger sample size will be needed to improve power for detecting these variants. However, it has been increasingly difficult to continue increasing the sample sizes of GWAS. Recently, substantial efforts have been devoted to improving the power by leveraging the functional information of genetic variants (e.g., gene expression) as well as the information from multiple related phenotypes. In this talk, we present a summary statistics-based method for testing genetic association with a set of variants, incorporating the information on genetic regulation of molecular characteristics. Here the summary statistics include the genome-wide marginal association regression coefficient estimates and standard errors, as well as the linkage disequilibrium of genetic variants. I will also show the connection to the popular Mendelian Randomization analysis. Simulation demonstrates that summary statistics-based p-values agree well with those from individual level data, but with substantively improved computational speed. Importantly, a broad application of our method to GWAS is possible, as only GWAS summary statistics are required.