Cascading Epigenomic Model for GWAS

Deciphering the mechanisms through which GWAS SNPs affect phenotypes is challenging. A number of methods, such as MetaXcan, have been put forth for associating the genetic component of either gene expression or DNA methylation to phenotypes. In this work, we propose a cascading epigenomic analysis for GWAS (CEWAS) to associate the epigenomic component of gene expression that is driven by genetics to phenotypes. On a well-powered GWAS, we show that CEWAS provides higher detection sensitivity than MetaXcan. Importantly, we show with simulations that statistics generated by CEWAS are properly calibrated. Further, using atSNP, we show that many SNPs associated with the detected genes affect transcription factor (TF) affinity. In fact, some of the detected genes are found to be potential TF targets, illustrating another utility of CEWAS.