Fig. 2

Results on simulated data. a Allele-specific copy number profiles from the ground truth as well as predictions by CNRein, SIGNALS, and CHISEL for a single simulation instance with a whole genome duplication at the trunk. b Accuracy values of unordered allele-specific copy numbers. c L1 errors of unordered allele-specific copy numbers. d The number of predicted unique copy number profiles (y-axis) vs. ground truth number of unique profiles (x-axis), with the \(y=x\) line in red and Pearson correlations indicated in the key and dashed lines. A small amount of jitter (drawn from \(\textrm{Unif}(-25,25)\)) was added to predictions (y-axis) with exactly 1000 unique copy number profiles for clarity. e Parsimony scores for phylogenetic trees inferred from ground truth (x-axis) and predicted copy-number profiles (y-axis), with the ground truth \(y=x\) line shown in red, and best-fit lines shown with dashes as well as Pearson correlations indicated in the key. The best-fit lines can appear slightly curved as a consequence of the plot’s log scale