Fig. 2
![Fig. 2](http://media.springernature.com/full/springer-static/image/art%3A10.1186%2Fs13059-018-1417-1/MediaObjects/13059_2018_1417_Fig2_HTML.gif)
Accuracy analysis with simulated data. a Proportion of events measured by each method (y-axis) from the 277 positive simulated cassette events at different sequencing depths (x-axis), from 120 million (120M) down to five million (5M) paired-end reads, using 100-nt paired-end reads. b As in a but for different read lengths (x-axis) at fixed depth (25 M). c True positive (TP) rate (in terms of percentage) for each method (y-axis) at different sequencing depths (x-axis) for 100-nt paired-end reads. TPs were calculated as the number of statistically significant events according to each method: corrected p value < 0.05 for SUPPA2, rMATS, and DEXSeq; and posterior(|ΔPSI| > 0.1) > 0.95 for MAJIQ. d As in c but for different read lengths (x-axis) at fixed depth (25 M)