Fig. 2

Relative error in estimating IUPM for data simulated under the standard experimental design. We calculated the relative error of an estimate \(\hat{x}\) given true value x as \(|(\hat{x}-x)/x|\). Each set of box-and-whisker plots summarizes the relative errors for estimates obtained by IUPMStats (red) and IUPMBayes under three different sets of variant frequencies (see inset legend) for a given true value of IUPM (0.2, 1, 5 and 25 per million cells). We used a log-transformation of relative errors and rescaled the y-axis to clarify differences between methods and simulation conditions; 25 outliers with relative errors below .03 were excluded from this plot region