Fig. 4

Pan-cancer pathway activation in non-T cell-inflamed relative to T cell-inflamed tumors from RNAseq gene expression. a Each tumor type shows transcriptional programs activated in non-T cell-inflamed relative to the T cell-inflamed tumor group. Thirty-one transcriptional programs present in four or more tumor types are shown. Red color labels those at activation z-score ≥ 1.95 predicted by IPA causal network analysis (see the “Methods” section). The full list of 266 pathways activated in one or more tumor types is provided in Additional file 1: Table S10. b External validation of activated transcriptional programs at gene expression level in three independent cancer genomic databases (ICGC, CPTAC, and MET500). Thirty-one transcriptional programs from a are shown on the row, and 13 tumor types with at least one validation cohort are shown on the column. Cohorts from TCGA or independent validation datasets are shown side by side for the same tumor type. The description and sample size of studies from each database are provided in Additional file 1: Table S1. Size of circle represents Pearson’s correlation coefficient r of each transcriptional program versus T cell-inflamed gene expression, with blue color indicating negative correlation (Pearson’s r < 0) and red color indicating positive correlation (Pearson’s r > 0). c RPPA validation of activated transcriptional programs at protein level. Twenty-five out of 31 transcriptional programs from a are shown. Six do not have corresponding protein in RPPA data, hence not shown (NE = not evaluable). Color represents Pearson’s correlation coefficient r of each transcriptional program versus T cell-inflamed gene expression, with blue color indicating negative correlation (Pearson’s r < 0), and red color indicating positive correlation (Pearson’s r > 0). The full list of correlation statistics is provided in Additional file 1: Table S11. d KLF4 pathway activation in a clinical cohort. Only pre-treatment samples are shown (n = 73). Left panel: correlation between KLF4 pathway gene expression and T cell-inflamed gene expression. Right panel: KLF4 pathway gene expression in non-responders (NR, including PD and SD) (n = 33) and responders (R, including PR and CR) (n = 40). IPA’s causal network analysis is used in a, Pearson’s correlation is used in b, c, and the left panel of d. Two-sided Welch Two Sample t-test was used in the right panel of d