Fig 2. Power of stratified GWAMA approaches to identify GxS for balanced strata design.

Shown is the power to detect GxS in equally sized strata (n₁ = 100,000, n₂ = 100,000) for each of the considered approaches, for varying effect sizes in stratum 2, $R_2^2$, with a fixed genetic effect in stratum 1, $R_1^2$, that is (A) small ($R_1^2=0.014\%$), (B) medium ($R_1^2=0.058\%$), or (C) large ($R_1^2=0.167\%$). The effect sizes for $R_1^2$ are chosen as those observed for WHRadjBMI near STAB1, PPARG or LYPLAL1, respectively. The modeled GxS are visualized on the left side (red bar: $R_1^2$, blue arrows: varying $R_2^2$). For the difference test without filtering, we assume a significance level at 5 x 10⁻⁸; for approaches with filtering, the filtering threshold is 1 x 10⁻⁵ and the significance level applied for the consecutive difference test is α_Diff = M/0.05, with M being the number of filtered lead variants (see Methods).