Fig. 1
From: Improving the performance of Bayesian phylogenetic inference under relaxed clock models
The models and prior distributions to simulate the sequence data. The sequence alignment (SA) is simulated through a phylogenetic continuous-time Markov Chain (PhyloCTMC) that consists of a substitution model (HKY) and an uncorrelated relaxed clock model (UCRelaxedClockModel). The random variables in HKY model construct the mutation rate matrix (Q), including base frequencies (π={πA,πC,πG,πT}) and kappa (κ). The time trees (ψ) and branch rates (ri for each branch i in ψ) construct the substitution tree (ST). The branch rates have a LogNormal prior with fixed mean 1 and certain standard deviation (denoted by s1). And the time trees have a Yule model prior with birth rate (λ) having a LogNormal prior. The other prior distributions include a Dirichlet distributions on π, a LogNormal distribution on κ, and a LogNormal distribution on s1. For notations in LogNormal distributions, the uppercase letters represent the parameters in real space, and the lowercase letters represent the parameters in log space. In all the simulations, the number of taxa is fixed at 120 (n = 120)