Fig. 3. Construction and validation of ratio-based transcriptome-wide reference datasets.

a, Workflow for constructing Quartet RNA reference datasets. Reference datasets were constructed according to the following steps: (1) identifying detectable genes; (2) calculating ratio-based expression based on reliably detectable genes that were differentially expressed; (3) assessing the homogeneity and stability of RNA reference materials; (4) assessing the uncertainty of ratio-based reference datasets; and (5) identifying high-confidence DEGs. b–d, Scatter plots of log₂ fold changes (FCs) of gene expression between reference DEGs and RT–qPCR (b), ddPCR (c) and proteomics data (d). Pearson correlation coefficient across three sample pairs was calculated. Genes/proteins that were considered as differentially expressed in both methods shown in x axis and y axis were used for plotting. x axis: average log₂FC from 13 high-quality RNA-seq batches reference DEGs. y axis: for the RT–qPCR and proteomics data, a gene or a protein was considered as a DEG/DEP when the t-test two-sided P < 0.05 and FC ≥2 or ≤0.5; for ddPCR data, genes that were identified as DEGs based on RT–qPCR were used. Average log₂FC of RT–qPCR (n = 3), ddPCR (n = 2) and proteomics (n = 3) data from DEGs/DEPs were used for plotting. DEP, differentially expressed protein.