Fig. 7

Construction and validation of the prognostic NRG_score. A Generation of the ten gene groups after 1000 iteration. The gene model with 13 genes was selected to construct the signature for NRG_score as its highest frequencies of 726 compared to other four gene models. B The c‑index of both training and testing sets. C Alluvial diagram of NRCs in groups with different gene clusters and NRG_score groups. D Barplots show the risk score between three NRCs and three gene clusters. The statistical difference of three clusters was compared through the Kruskal–Wallis H test. *P < 0.05; **P < 0.01; ***P < 0.001. E Ranked dot and scatter plots showing the NRG_score distribution and patient survival status. F, G Kaplan–Meier analysis of the survival rate between the two groups. The high and low groups were divided by the median value of the NRG_score (left pannael). ROC curves to predict the sensitivity and specificity of 1-, 2-, 3-, and 5-year survival according to the NRG_score (right panel). H Barplot shows the NRG_score between groups with adjuvant chemotherapy (ADJC) and without adjuvant chemotherapy (ADJC). The statistical difference of two clusters was compared through the Wilcox test. *P < 0.05; **P < 0.01; ***P < 0.001. I Survival analysis among four patient groups stratified by both NRG_score and treatment with adjuvant chemotherapy (ADJC). J, K Differences in the expression of 33 NRGs and 13 genes among the two gene subtypes. The statistical difference of two groups was compared through the Wilcox test. *P < 0.05; **P < 0.01; ***P < 0.001