Gene-environment interaction in genome-wide association studies

Abstract

It is a commonly held belief that most complex diseases (e.g., diabetes, asthma, cancer) are affected in part by interactions between genes and environmental factors. However, investigators conducting genome-wide association studies typically test for only the marginal effects of each genetic marker on disease. In this paper, the authors propose an efficient and easily implemented 2-step analysis of genome-wide association study data aimed at identifying genes involved in a gene-environment interaction. The procedure complements screening for marginal genetic effects and thus has the potential to uncover new genetic signals that have not been identified previously.

Figure 1.

Power for 1-step and 2-step analyses for varying levels of interaction effect size (R_ge). All other parameter settings remain constant under “base” model specifications (M = 10,000, number of cases/controls = 500/500, q_A = 0.2, p_e = 0.5, R_g = 1, R_e = 1, p_ge = 0, α₁ = 0.05).

Figure 2.

Percentage of replicates for which the P value for disease susceptibility locus is ranked in the top k (k = 10 or k = 25) marker P values for varying levels of interaction effect size (R_ge). All other parameter settings remain constant under “base” model specifications (M = 10,000, number of cases/controls = 500/500, q_A = 0.2, p_e = 0.5, R_g = 1, R_e = 1, p_ge = 0, α₁ = 0.05).