We developed a simulation framework to evaluate the power of MAGENTA to identify enrichment of multiple associations for which we have low detection power with single SNP analysis. SNPs with a small effect size were randomly spiked into varying numbers of genes (referred to as causal genes) in pre-specified gene sets (one SNP per gene), and into genes outside the gene set, maintaining the total number of causal genes in the genome. The simulations were performed on a background of randomized SNP association p-values, for all SNPs i in the genome, generated with phenotype permutations of the DGI study (see section above). For each set of parameters tested, 1,000 simulation runs were performed. In each simulation run, the genes representing a simulated gene set of a given size were randomly chosen from the genome, and the various fractions of genes assigned a SNP of small effect size were also randomly chosen from all genes in the gene set. The remaining number of causal SNPs was randomly assigned to genes outside the gene set. The small effects were randomly assigned to
