a, Relative ordering model (PhylogicNDT LeagueModel) results for a simulated cohort of samples (n = 100) from a single generalized relative order of events (with varied prevalence) showing high concordance with the true trajectory. Probability distributions show the uncertainty of timing for specific events in the cohort. b, Relative ordering model results on a simulated cohort of samples (n = 95) from a complex mixture of trajectories with different order of events showing high concordance with the expected average trajectory. c, Estimation of accuracy of the relative ordering model reconstruction by simulation of a set of 100 cohorts (n(samples) = 100) with random trajectory mixtures and quantifying the distance in log odds early/late from perfect ordering. For the vast majority of events (even with low number of occurrences in the cohort), the log odds error does not exceed 1, confirming that very few events would switch between timing categories. The inset box corresponds to the first and third quartiles of the distribution, the horizontal line indicates the median and whiskers include data within 1.5 the IQR from the box. d, Simulated data show concordant timing in cohorts with WGD (n = 245). Exclusion of samples with WGD (right, n = 242) introduces only a mild drop in accuracy, indicating that WGD is beneficial but not necessary for the reconstruction. Red dot = true rank. e, Estimated log odds in observed data including WGD (left, n = 245) and without (right, n = 242), across different mutation types. The inset box corresponds to the first and third quartiles of the distribution, the horizontal line indicates the median and whiskers include data within 1.5 the IQR from the box.
perfect timing catherine anderson ebook 28
2ff7e9595c
Comments