Fig. 3.

Power in in silico experiments and simulations. a True positive rate (y-axis) for increasing α-level cutoffs (x-axis) in the yeast RNA-seq in silico resampling experiment with spiked-in differentially expressed genes (left panel) and the simulation of yeast RNA-seq counts using the polyester R/Bioconductor package [28]. Similar plots in the style of [31] are displayed in Additional file 1: Figure S2. b True positive rate (y-axis) across simulation settings at α-level of 0.05. The left panel displays increasing numbers of hypothesis tests, and the right panel displays increasing proportions of non-null hypotheses. Note that the lfdr method is displayed as a dotted line when the number of tests per bin falls below 200 (where the number of bins is fixed at 20), as fdrtools generates a warning in this case that the estimation may be unreliable