Fig. 5: Contribution of cancer splicing programs to disease progression.

a Cancer splicing program regulation during carcinogenesis. Carcinogenic transformation begins with genomic lesions that activate oncogenes and inactivate tumor suppressors, reprogramming the cell into a cancerous state. b ORA of the top 10 enriched ReactomeDB terms for genes corresponding to target exons of splicing factors in the oncogenic-like and tumor suppressor-like cancer splicing programs. Enriched terms only in the top 10 for one of the splicing programs are also shown. c Prioritization of exons potentially contributing to ICB response by integrating differential splicing between responders and non-responders (p value < 0.05, two-sided Wilcoxon rank sum test) with orthologous gene-level immune evasion scores from CRISPR screens in immunocompetent ICB-treated versus immunodeficient mice. Color indicates exon impact (VastDB). d Kaplan Meier survival curves of ICB patients in Riaz et al.⁴⁶ stratified according to the HsaEX1036341_SEC22B inclusion levels that maximized log-rank test differences. Color, patients with high and low exon inclusion. Bottom, exon inclusion PSI cut point, and two-sided log-rank test p value. Shaded bands indicate 95% confidence intervals. e Median activities of cancer splicing programs correlate with cancer cell proliferative state. Top, Two-sided Pearson correlations (R) and corresponding p values (p). Error shadows, 95% confidence intervals. For the left panel, the exact p-value underflows to 0 due to numerical precision limits and is therefore reported as p < 2.2 × 10^-16. f Danielsson et al.⁴⁹ experimental design. Primary fibroblasts were induced into different states by transfecting either hTERT (immortalized), hTERT and SV40 large-T (tumorigenic), hTERT, SV40 large-T, and RAS G12V (metastasizing). g Splicing factor activities by cancer splicing program in different stages of carcinogenesis. Top, p value based on two-sided Wilcoxon rank sum tests comparing the activity of the splicing factors in each cancer program. n, number of splicing factors with activity estimates available for each program. In box and whisker plots in panel (g), the median is indicated by a horizontal line, boxes represent the first and third quartiles, whiskers extend to 1.5 times the interquartile range, and outliers are shown individually. Source data are provided as a Source Data file.