Fig. 6

Accuracy on simulated data. The two panels (a) and (b) show the high correlation of true and inferred maximum likelihood tumor purity and ploidy in simulated data. Colors visualize the simulated purity, ranging from 20 to 80%. Excluding the samples with tumor purity of 20%, the correlations increase to 0.98 for purity and 0.95 for ploidy [(c) and (d)]. Note that for almost all of the samples, the correct solution was considered, but was not always ranked as maximum likelihood solution. Panel (e) shows for all samples, including the 20% samples, the ploidy correlation of the best considered solution (minimizing the Euclidean distance in scaled purity and ploidy), for example selected in a hypothetical perfect manual curation (Pearson 0.98). Panel (f) shows the histogram of ranks of the best solution over all samples