Genome-wide meta-regression of gene-environment interaction

Understanding the effects of gene-environment interaction on complex human diseases or traits in genome-wide association studies (GWAS) can help uncover novel genes and identify environmental hazards that influence only certain genetically susceptible groups. Thus there is a pressing need to develop efficient and powerful interaction analysis methods. In this paper, we propose a novel meta-analysis method of gene-environment interaction, based on meta-regression (MR-M&I). Compared with existing meta-analysis methods, MR-M&I allows for heterogeneity in the environmental factor (E) by dividing the subjects in each study into groups according to the distribution of E. Moreover, it can readily estimate linear or non-linear interactions, and thus it is more generally applicable to different scenarios. We use numerical examples to demonstrate the performance of MR-M&I and compare it with two commonly used methods in current GWAS. The results show that MR-M&I is more powerful than the other methods.