Supplementary Figure 6: Tumor-suppressor gene candidates examined for mechanisms of inactivation via NMD and copy number changes.

(a) Principal-components plot as in Figure 5g, with nine additional genes overlaid (blue points). The additional genes were proposed to be possible tumor suppressors in Davoli et al.³⁷, on the basis of their enrichment for deleterious over benign somatic mutations (the TUSON Explorer method, FDR ≤ 20% for all except ZFHX3 (22%) and MKI67 (31%)). Examined genes have ≥50 nonsense mutations in our data set. (b) Relative frequencies of NMD-inducing versus NMD-evading nonsense mutations (“NMD” and “noNMD”, respectively) and of gene deletions versus absence thereof (“Del” and “noDel”, respectively). Importantly, in the predictor from Davoli et al., nonsense mutations were not subclassified by their putative effects on NMD nor were copy number alterations used as input.