Fig. 2

Comparison of the usefulness of NATs and tumor tissues for discovering prognostic gene markers. A–C Two types of DEGs, NAT-DEGs and tumor-DEGs, were identified by comparing gene expression levels between RC and nonRC samples using the NAT-derived transcriptomes and tumor-derived transcriptomes. A QQ plots comparing P-values of NAT-DEGs and tumor-DEGs. B Bar graphs showing the number of NAT-DEGs and tumor-DEGs obtained with three FDR cutoffs (0.0001, 0.001, and 0.01). C Volcano plots generated by selecting NAT-DEGs (left) and tumor-DEGs (right) based on the thresholds of − log₁₀(FDR) > 2. The names of selected DEGs identified with these thresholds are shown. Yellow shaded areas represent − log₁₀(FDR) > 2 and abs(log₂FC) > 2. D Comparison of the − log₁₀(FDR) (upper panel) and log₂FC (lower panel) of the changes in gene expression between RC and nonRC states of 51 previously identified prognostic marker genes in NATs and tumors. Genes with lower FDR values in NAT than in tumor are on the left panel, and vice versa on the right panel. Note that P-values rather than FDR values were assigned to SLC6A11, in either NAT-derived transcriptome or tumor-derived transcriptome, because due to too low mean counts for this gene resulting in ‘NA’ by DESeq2. The grey bars on the lower panels represent the genes with no statistical difference between NAT and tumors (FDR > 0.05). N.S., not significant. E QQ plot of P-values estimated from (D) in NATs and tumors