Model-free or machine learning approaches in the context of GWAS are relatively new and currently computationally expensive. The performance of each method will vary with the distributional assumptions underlying the phenotypic outcome, the environment and their suspected interaction. The Analysis Subcommittee considers each approach and its strengths, limitations and feasibility for a particular scenario, and advises SI on the most appropriate method for their G × E interaction of interest. Each SI is responsible for data analysis and plans to replicate initial findings as outlined in their response to the RFA. G × E interactions will also be investigated in cross-study trait analysis; some of which are sufficiently powered even when applying a conservative test for interaction. For example, gene-smoking interactions for both BMI and caffeine consumption are highly anticipated and we will have 80% power to detect marginal R2 for interaction effects as modest as 0.0013 and 0.002, respectively.
