Extended Data Fig. 4: Definition of tumor purity level and characteristics of each SP in the discovery cohort.

a, Heatmap showing the hierarchical clustering of CT-based-CNs and the percentage of CT-based-CNs. The right panel showed annotations of defined Tumor Purity Level (TPL). The Kaplan–Meier survival curves of TPL were drawn and log-rank tests were done to investigate the significance. L, n = 95; M, n = 105; H, n = 88. b, Representative images of p53 expression in tumor cells. c, Dot plot showing the percentage of each CP-based-CN in distinct SPs. The value and size of the dot represented the percentage. *p represented it was significantly higher or lower than each other SP. Two-sided Wilcox tests. *p < 0.05. d, Dot plot showing the z-scored percentage of each cell type in distinct SPs. The value and size of the dot represented the z-scored percentage. *p represented it was significantly higher or lower than each other SP. Two-sided Wilcox tests. *p < 0.05. e, The Kaplan–Meier survival curves of SPs in samples of TNM stage I&II and III&IV respectively. Log-rank tests. For analysis of TNM stage I & II: SP-HI, n = 48; SP-LI, n = 64; SP-Endo, n = 26; SP-TIM, n = 44; SP-PF, n = 33. For analysis of TNM stage III & IV: SP-HI, n = 18; SP-LI, n = 23; SP-Endo, n = 9; SP-TIM, n = 13; SP-PF, n = 10. f, The Kaplan–Meier survival curves of each SP vs all other samples. Log-rank tests. n = 288. SP-HI, n = 66; SP-LI, n = 87; SP-Endo, n = 35; SP-TIM, n = 57; SP-PF, n = 43. The exact p value in c, d can be found in Supplementary Table 3. For all panels, n denotes biological independent samples.