Fig. 4
From: Impact of adaptive filtering on power and false discovery rate in RNA-seq experiments
Adaptive filter II. Differences in power for adaptive filter and selection of applied filters compared to no filter for NB sim distributed for Kidney and Sultan data, SimSeq simulation for Bottomly and mouse mammary data sets and PROPER simulation for sequencing depths 5m and 50m for varying \(\pi _0\), \(m=10000\), \(n_1=n_2=10\) (\(n_1=n_2=5\) for the SimSeq Bottomly data and \(n_1=n_2=3\) for SimSeq mouse data) and \(\alpha =0.05\) (lfdr adjustment). The plotted filtering methods and corresponding percentile percentages are given in the legend. Note that the range of the y-axis is chosen result-based; filtering methods with low power may not be visible on some plots