Fig. 2

Analysis of a hepatocellular carcinoma sample with PCAWG consensus calls. a CNAqc visualization: allele-specific CNAs (ploidy 2, purity ~ 85%) with major and minor allele counts per segment. This sample harbors two driver SNVs hitting genes CTNNB1 and ALB, sitting in diploid heterozygous segments (1:1). b–d Read counts for SNVs visualized as variant allele frequencies (VAFs) and depth of sequencing (DP). Cancer cell fractions (CCF) obtained by CNAqc suggest that the two drivers are clonal (CCF spread around 1). e Peak detection QC for simple clonal CNAs, as in Fig. 1f. Peaks are checked independently, and the final QC depends on the number of mutations per peak, and whether the peak is matched. The sample-level QC is a linear combination of results from each CNA; here calls are passed (green plot; numbers represent mutational burden). f,g CCF estimation for mutations mapping to triploid 2:1 segments, obtained using the entropy-based and the rough methods. CCF values of clonal mutations spread around 1, CCFs and VAFs are colored by mutation multiplicity. The entropy profile (dashed line) delineates crossings of binomial densities where CNAqc detects multiplicity uncertainty from VAFs; the entropy method detects uncertainty in 20% of the SNVs. The alternative method in panel g assigns multiplicities regardless of entropy. In both cases, the CNAqc CCF estimates pass QC with default parameters