Fig. 4From: Gene dispersion is the key determinant of the read count bias in differential expression analysis of RNA-seq dataComparison of read count bias for three different replicate type datasets. For ten published RNA-seq datasets, a the boxplots of SNRs are depicted against the read counts. Each bin contains 1000 genes. Each row of figures represents technical (MAQC-2, Marioni, Oliver dataset; pink), GI (Liu, Li and Barutcu dataset; green), and unrelated (TCGA BRCA, KIRC, PRAD and Tuch dataset; blue) replicate data, respectively. b The proportions of DE genes in each bin were plotted for each dataset. The DE genes were obtained by using the R packages baySeq, DESeq, DESeq2, edgeR, MASS (naïve LRT) and SAMseqBack to article page