Fig. 1

a Best-fitted distribution to discover feature genes. Cullen Frey method was applied to identify the best distribution fitting for mutational data of cancer types. The figure shows the Cullen-Frey graph for discrete distributions. b The distribution of mutated genes in cancer samples. We fitted a negative binomial to their mutational data for all cancer types. Each plot shows empirical mutation data in a specific cancer type, and the red line shows a negative binomial distribution fitted to the cancer type. The X-axis indicates the number of mutated genes, and Y-axis shows the fraction of samples in the specific cancer type. If a bar in x = 120 has y = 0.02, then there are 120 genes with mutations in 0.02 of samples of that cancer type. The distribution’s right tail points to mutated genes in more samples, therefore, more important genes