Fig. 9
From: Stability of methods for differential expression analysis of RNA-seq data
Impact of factors on stability. a AUCOR against number of samples for each condition that are 2,3,4,5,8,11,14,…. b AUCOR against fold change. Fold changes of DE features are generated from the normal distribution with standard error 0.5 and the means of fold changes are set as 2,2.5,3,…,6. c AUCOR against dispersion. Basic pairs of mean and dispersion are randomly selected from that of Pickrell data [25]16. Dispersions are adjusted by multiplying a ratio from 0.6 to 2 with step size 0.1. d AUCOR against proportion of DE features that are up-regulated. e AUCOR against number of features. f AUCOR against threshold. Features with adjusted p-values less than the threshold are identified as DE features. We consider 5 commonly used thresholds: 0.001,0.01,0.05,0.1,0.2. g AUCOR against proportion of DE features that is spread from 10% to 70%. h AUCOR against proportion of outliers. i AUCOR against outlier mechanisms: N, S, R and M. N represents the case without outliers. Different DE methods are represented by different symbols and colors