Second, at the present time, we sample each extant lineage with probability is fixed, so we may suppress it in our notation. Pruning all dead and nonsampled lineages, as well as the lineage ancestral to the first branching event yields the “sampled tree,” for the sampled lineages only (Fig. The sampled tree is drawn such that each branching event has the “left” descendant on the left and the “right” descendant on the right.The distinction between left and right is a convenient notation for the derivation of the probability density.[Bayesian inference; MCMC; molecular clock dating; sampled tips; viral evolution.] The distance information in molecular sequences can be translated into absolute times and rates if information about the ages of some nodes in the phylogeny is available.This strategy has been used to date species divergences, with the fossil record used to inform the ages of certain nodes and thus to calibrate the molecular phylogeny. 1998) provides a powerful general framework for integrating such different sources of information.Third, we also condition on the topology of the sampled tree (i.e., on the sampled tree ignoring branching times and sampling times, but preserving relatedness and orientation of “left” and “right”).
This type of data is commonly collected during viral epidemics and is sometimes available from different species in ancient DNA studies.Although the pruning algorithm for likelihood calculation (Felsenstein 1981) is implemented in BEAST, MCMCtree in addition implements an approximate method for the likelihood calculation (Thorne et al.1998; dos Reis and Yang 2011), which is much faster and can be used in analysis of large data sets (Battistuzzi et al. We extend the birth–death-sequential-sampling (BDSS) model of Stadler (2010) to specify a prior distribution of divergence times, which is combined with the prior on the evolutionary rates and with the likelihood of the sequence data to give the posterior distribution of divergence times.The BDSS prior is very flexible and, with different parameters, can generate trees of very different shapes, suitable for examining the sensitivity of posterior time estimates.We apply the method to a data set of SIV/HIV-2 genes in comparison with a likelihood-based dating method, and to a data set of influenza H1 genes from different hosts in comparison with the Bayesian program BEAST.