Fig. 1

FANGS Flow Chart. A specific dataset is selected as baseline. The raw data is preprocessed as described in the Methods section, which includes RMA and quantile normalization along with centering the data for each gene to have mean zero, to remove the existing signal. Next, a gene set is selected for simulation, and a range of association signals are introduced, with a range of effect sizes, and proportions of the genes in the gene set that are associated with disease status. One hundred bootstrapped data sets are generated for each set of simulation parameters. The simulated data is tested on each method and power to detect the differentially expressed gene set is calculated