A0497
Title: Nonlinear moderated mediation analysis with genetical genomics data
Authors: Yuehua Cui - Michigan State University (United States) [presenting]
Abstract: Genetical genomics data provide promising opportunities for integrative analysis of gene expression and genotype data. Here we propose a nonlinear moderated mediation analysis taking into the causal mediation effect of gene expression on the relationship between genetic and phenotype, potentially moderated by environmental factors. Our goal is to select important genetic and gene expression variables that can predict a phenotypic response under a high-dimensional setup. As genes function in networks to fulfill their joint task, incorporating network or graph structures in a regression model can further improve gene selection performance. We propose a nonlinear graph-constrained penalized regression model to improve the selection performance via incorporating gene network structures. A two-step estimation procedure is adopted to obtain better variable selection and estimation. Simulation and real data analysis are conduced to show the utility of the method.
