Empirical, simulation-based statistical power of GGG tests.

aIndices of the interaction SNPs in the two loci (Figure S3); Three types of scenarios are considered, of one, three, and five pairs of interacting SNPs, for each at least two different sets of SNPs are considered, for a total of 7 different scenarios.
bU: SNP untyped; O: SNP observed; For scenarios with more than one pair of interacting SNPs, the U/O status of the first and second interacting SNP is the same across all pairs.
cMinor allele frequency of all SNPs involved in interactions, by order.
dCoefficient of the interaction term in the linear model, b3, as described in Equation (1); For scenarios with more than one pair of interacting SNPs, the effect size is the same for all pairs.
ePower, as percentage of significant tests with P value<0.05.