Fig. 4.

Relative errors in IUPM estimates for simulated data with uniform cell counts. Each box plot summarizes the distribution of relative error for 100 replicate simulations for the respective methods: IUPMStats, QVOA-NGSA, and IUPMBayes (three sets of variant frequencies, see legend). The plots are drawn on a log-transformed y-axis to facilitate comparison between methods under varying conditions. When the maximum likelihood estimators employed by IUPMStats or QVOA-NGSA were unable to produce a finite estimate of IUPM, we assigned an arbitrary value of 100 cells per million. The respective box plots are directly labeled with the numbers of non-finite estimates at the outliers or median bands