Fig. 2From: BANDITS: Bayesian differential splicing accounting for sample-to-sample variability and mapping uncertaintyTPR vs. FDR for transcript-level testing for the three 2-group comparison simulation studies. a 3 vs. 3 simulation study. b 6 vs. 6 simulation study. c 6 vs. 6 simulation study with transcript pre-filtering (transcripts with at least 10 counts and an average relative abundance of 0.01). Circles indicate observed FDR for 0.01, 0.05, and 0.1 significance thresholds. Note that for cjBitSeq, we considered the probability that a transcript is not differentially used, which does not guarantee FDR controlBack to article page