Fig. 4: Splicing factors recurrently active in tumors exhibit cancer driver-like behavior. | Nature Communications

Fig. 4: Splicing factors recurrently active in tumors exhibit cancer driver-like behavior.

From: Exon inclusion signatures enable accurate estimation of splicing factor activity

Fig. 4: Splicing factors recurrently active in tumors exhibit cancer driver-like behavior.The alternative text for this image may have been generated using AI.

a TCGA cancer cohorts overview. Cancer type abbreviations are available at (https://gdc.cancer.gov/resources-tcga-users/tcga-code-tables/tcga-study-abbreviations). b Workflow to identify tumor-specific splicing programs. From healthy and tumor samples, we quantified their exon inclusion and computed the splicing signatures to obtain splicing activities for each splicing factor and sample. c Frequency of splicing factors showing significant activation (top) or inactivation (bottom) in tumor samples versus solid tissue normal (FDR < 0.05, two-sided Wilcoxon rank sum tests). Dashed lines mark the recurrence threshold for a splicing factor’s inclusion in either the oncogenic-like or tumor suppressor-like cancer splicing program, with selected splicing factors above each line. Labels, top 5 most recurrently activated or inactivated splicing factors. d Workflow to associate protein activities in primary tumor samples with patient overall survival using a Cox Proportional Hazards (Cox P.H.) survival model. Positive Cox P.H. coefficients indicate that high splicing factor activity is associated with a high risk of death. Conversely, negative Cox P.H. coefficients indicate that high splicing factor activity is associated with a low risk of death. e Number of cancer cohorts in which either CPSF6 (left) or DNMT1 (right) is associated with either high or low risk of death, respectively. Top, number of cancer cohorts for each type of association. Thick border, number of cancer cohorts with two-sided Cox Proportional Hazards coefficient significantly different than 0 (FDR < 0.05). f Distributions of the number of cancer cohorts for which each splicing factor is associated with a high or low risk of death. Top, p value based on two-sided Wilcoxon rank sum tests. Bottom, number of splicing factors in the corresponding cancer splicing program. g, h Splicing factors from cancer splicing programs identified in (g) the spliceosome reaction complexes according to SpliceosomeDB43, and in (h) RNA-binding protein (RBP) domain families. In box and whisker plots in panel (f), 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.

Back to article page