Abstract
Although both genetic and environmental factors play important roles in the etiology of many complex human traits, scans for gene(G)-environment(E) interaction have been mostly neglected in current GWAS. This is mainly due to the substantial lack of power for interaction tests coupled with the massive number of interactions available for testing. The conventional statistical approach for interactions, the case-control analysis, suffers from low power. The alternative case-only analysis can be more powerful but slight violations in the underlying assumption of independence can greatly bias the results. In this talk, I discuss several recently proposed alternatives ranging from two-step analyses to Bayesian approaches. Each approach attempts to leverage various aspects of the data, from ascertainment, to marginal test, to estimates of G-E association in the controls. For example, in the two-step approach proposed by Murcray et al. (2009), a preliminary screening step tests the association of G with E in the combined sample of cases and controls, thus capturing any G-E association that may be present in the study sample by leveraging the oversampling of cases from the population. This step is followed by a conventional case-control analysis. Overall, there is a substantial increase in power compared to a genome-wide scan of interactions due to the drastic reduction in the number of tests evaluated. In contrast, several Bayesian approaches (Mukherjee and Chatterjee, 2008; Li and Conti 2009) attempt to obtain an estimate of interaction that is a combined estimate between the case-only and case-control estimates. While originally conceived as an approach for a single GxE interaction test, I will discuss recent extensions leveraging the fact that most SNPs in a GWAS are not associated with either the outcome nor the environmental factor. For these and several other approaches, I will discuss the underlying assumptions and the sensitivity of the performance of each to G-E association in the population, population prevalence, and the assumed causal model.
Citation Format: David V. Conti. Searching for GxE interactions in genome-wide association studies. [abstract]. In: Proceedings of the AACR Special Conference on Post-GWAS Horizons in Molecular Epidemiology: Digging Deeper into the Environment; 2012 Nov 11-14; Hollywood, FL. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2012;21(11 Suppl):Abstract nr IA04.