Fig. 8

Model comparison highlighting the importance of indirectly acting gene CNAs for survival analysis. Comparison of our standard survival impact quantification (CCTN) to a basic version (CCTN: direct neighbors only). The standard approach considers all gene CNAs that directly or indirectly impact on survival signature genes, whereas the basic version accounts only for directly acting gene CNAs to realize a classification into short- and long-lived patients. a Percentage of patients per cohort that can be classified by both approaches. CCTN reaches significantly increased numbers of classifiable patients compared to the basic version (p<2.2×10⁻¹⁶, Fisher’s exact test). b Separation quality of patients classified as short- or long-lived by both approaches. Corresponding p values quantify the separation between patients classified as short- or long-lived with respect to differences of the short- and long-lived Kaplan–Meier curve obtained for each approach under consideration of all tumor-specific gene CNAs. The y-axis is plotted in negative logarithmic scale. CCTN reaches a clearly improved patient separation for the majority of cohorts in comparison to the basic version. c Kaplan–Meier curves for separation of CLCGP patients into short and long survival based on frequent gene CNAs. CCTN reaches a significantly improved patient separation. d Like (c), but here for SKCM patients considering all patient-specific gene CNAs. CNA copy number alteration, CCTN cancer cell transcriptional regulatory network, HNSC head and neck squamous cell carcinoma, LUAD lung adenocarcinoma, SKCM skin cutaneous melanoma