Fig. 5

Summary metrics computed to rate methods for final recommendations. Several metrics were computed over all settings of the simulations, and yeast experiments, as well as all datasets and covariates in the case studies to evaluate the a FDR control, b power, and c, d consistency of the evaluated methods. In a and b, color denotes whether the metric is computed over simulations (Sims) or yeast in silico experiments (Yeast). In c and d, color denotes whether the metric is computed for TPR or FDR in the simulations (“Sims (TPR)” and “Sims (FDR)”, respectively), or for the percentage of rejections in the case studies “CS (%reject).” In all panels, the methods are shown on the x-axis, and the methods with superior performance are those with a low value of the y-axis metric. Cutoffs used with the metrics shown are provided in the “Methods” section