Fig. 2

CNI and functional CNA levels at gene level

A. The spectrum of CNI across 14 TCGA cancer types using raw bulk RNA-seq expression data and absolute CNAs. The model was applied on the TCGA OV, LUSC, LUAD, STAD, BRCA, BLCA, HNSC, UCEC, SKCM, GBM, LIHC, LGG, PRAD, and KIRC datasets, allowing the quantification of CNI in each. The dashed line shows the median CNI. The results indicate that OV lands on the high extreme of the CNI, along with LUSC, LUAD, STAD, and BRCA, while KIRC, GBM, SKCM, and LGG are located in the lower tail of the CNI spectrum. Nine of these 14 cancer types have been shown to have a similar trend in CNA abundance in previous literature, suggesting the abundance correlates with but does not equate CNA driverness. B. CNA transition point captured from nonlinear model curve as functional CNA level for CCNE1 as an example gene. C. The obtained functional CNA status (red target) as the point with the highest impact over gene expression and its range of variability (cyan bar) across the DECIDER cohort are visualized for some example genes including ATR, BRCA1, CCNE1, FOXA1, KRAS, MET, RAD52, RB1, ZNF195, and ZNF733P