Fig. 2

Iterative inference of the mean fitness. Comparison of the true mean fitness \(\bar{s}(t)\) with the mean fitness inferred by both FitMut1 and FitMut2, for different sequencing depths (columns) and \(\mu (s)\) (rows). Each row in the \(4\times 4\) array corresponds to one simulation of the evolution, with the columns differing by the average simulated sequencing depth per time point per lineage. The 5th column shows the DFE \(\mu (s)\) used in each simulation condition, and the prior \(\tilde{\mu }(s)\) used for both FitMut2 and FitMut1 across all simulated DFEs