Fig. 8: Impact of each bioinformatics step on the accuracy of DEG identification.

Using the benchmark datasets from 13 laboratories, the Matthews correlation coefficients (MCC) for 140 bioinformatics pipelines were measured based on the Quartet reference datasets, the Quartet TaqMan datasets, and the MAQC TaqMan datasets. The impact of each bioinformatics step on the accuracy of DEGs was assessed, including (a) different annotations, (b) different alignment tools, (c) different gene quantification tools, and (d) different DEG identification tools. Box plots indicate MCC values for 13 benchmark datasets when different annotations, alignment tools, quantification tools, and DEG identification tools are used, and data are presented as median values (center lines) and the upper and lower quartiles (box limits). e The AUC was analyzed for five DEG identification tools using RNA-seq data from lab01, and the AUC values for other high-quality benchmark datasets are displayed in Supplementary Fig. 57. AUC Area Under the receiver operating characteristic Curve.