Fig. 2
From: Impact of adaptive filtering on power and false discovery rate in RNA-seq experiments
Power comparison of different filters and sequencing depths. Power values for several filtering methods for PROPER simulation for \(\alpha =0.05\), \(\pi _0 =0.8\), \(m=10000\), \(n_1=n_2=10\) and sequencing depths 5m, 10m, and 50m. The power of each filtering method is plotted as a function of the actual mean proportion of filtered genes across all simulation runs for the set of genes with at least one non-zero count; only the proportion of the basic filter is based on the total number of hypotheses m. The basic, Jaccard and no filter results are represented by a point because these methods are based on a fixed threshold