Fig. 4

Validation of CNAqc with single-cell low-pass DNA sequencing. a Pseudo-bulk CNAs from single-cell low-pass data of an ovarian cell line [31]. This profile is obtained from pooling CNAs across several diploid tumor subclones. This artificial bulk with known ground truth is then used to validate CNAqc (Additional File 1: Fig. S9). b QC of LOH and diploid heterozygous segments from panel a, using true tumor purity (100%). c Quality control of complex 3:0 segments (cluster I; Additional File 1: Fig. S9). d Correlation between purity adjustments suggested by CNAqc and errors created artificially for single-cell data. By construction, the desired correction sits on the diagonal because the true purity is 100%; the tool achieves \({R}^{2}=0.88\), correlation test p-value \(p<1{0}^{-16}\). e Advanced QC of subclonal CNAs from admixing of tumor subclones with mirrored allelic imbalance, as originally detected in [32]. In this test, two triploid subclones with mirrored alleles (AAB versus ABB) are admixed. CNAqc can validates these calls and identify the true branching patterns of evolution (AB \(\to\) AAB | ABB), which is characterized by a peak of shared mutations at around 50% VAF