#22 · Systems biology

Quantifying topology, calibration, and model uncertainty in phylogenetic inference

The common thread is the evaluation of assumptions built into phylogenetic and phylodynamic inference, and of the support values reported from them. A benchmark of fixed-tree-topology strategies on viral datasets finds substitution-model parameters robust, contrasted with clock rates and node ages. Adjusted pairwise composite likelihood offers faster Bayesian node dating than full-likelihood MCMC and greater robustness to fossil calibration uncertainty. Net Bootstrap Support provides efficient confidence limits that flag phylogenetic signal heterogeneity and overconfidence in phylogenomic relationships. Bayesian model averaging over parametric coalescent growth models addresses model choice for phylodynamic inference. What is new is treating topology, calibration, model selection, and support as explicit objects of uncertainty quantification.