Figure 1  Directed Acyclic Graph Showing Gene-Environment Dependence by Mediation
Boxes are observed variables, circles are unobserved, and arrows indicate directions of causal relationships. Abbreviations are as follows: D, causal variant; M, marker genotype; X, environmental exposure; Y, outcome such as disease; A, composite variable for ancestry, which gives rise to correlation (LD) between D and M; and G, composite variable for common causes of X and Y, which may include additional genes. X mediates the effect of D on Y. For example, a CYP3A variant (D) affects the risk of breast cancer (Y) via its effect on age at menarche (X).