Fig. 1.

ROC curves of different methods under various levels of background noise. ROC curves of DESeq2, edgeR, XBSeq, XBSeq2 under low, intermediate or high level of background noise (a); ROC curves of different methods but only with highly expressed genes (genes above 75% quantile of expression intensity) (b); ROC curves of different methods but only with genes expressed at low levels (genes below 25% quantile of expression intensity) (c); Simulations were carried out 100 times and average AUC were used. Dataset with 3 number of replicates per condition, 10% DEGs with 1.5-fold change was used