Divergence time estimates using Beast v1.80 and CIPRES: A tutorial

Post date: May 22, 2014 11:25:04 AM

Phylogenetic divergence time estimates is one of the key tools when investigating different aspects in the evolutionary history of lineages. Its applications range from the dynamics of viral disease transmissions (e.g. influenza virus) to the estimation of divergence time among taxa. In this post I will start by provinding a brief history on phylogenetic dating estimation followed by an application on how to use the program Beast v1.80 using CIPRES Science Gateway.

Zuckerkandl & Pauling (1962) were the first to propose that time of divergence between two taxa (when two particular lineages separated from each other in the past?) can be estimated by the number of differences in two molecular sequences. The main assumption in their method was that the rate of molecular change was constant through time, and envisioned a molecular clock which could provide non-fossil independent estimates of speciation events. In essence they proposed a single, global rate of substitution. Later studies however have found that the rate of sequence evolution was not constant through time, and that new methods were needed to incorporate the fact that molecular lineages vary through time and across lineages.

The first generation of methods to overcome this issue were known as local and relaxed clocks, where different clocks were included in the phylogeny to account for different rates of molecular evolution. In 1997, Sanderson proposed a method that minimized the degree of change along branches using a rate-smoothing algorithm within a Maximum likelihood framework that become popular. An alternative approach also in a maximum-likelihood framework was subsequently developed by the same author (Sanderson, 1997). It has become clear these days that divergence time estimates based on molecular sequences can only provide a relative time scale to a phylogeny and that fossil data are a necessary component in the divergence dating methodology for time calibration in order to convert it to absolute divergence time. In fact it is argued now by many that multiple fossil calibration points are a requirement for accurate divergence time estimations.

One of the most popular programs to estimate phylogenies in a Bayesian framework is BEAST. It is also used in coalescent-based population genetics, and to study evolving populations as documented in ancient DNA or viral sequences. BEAST is written in the programming language of Java, which is multi-platform and thus can be executed in any computer platform. The program is aimed to infer phylogenies using strict or relaxed molecular clocks.

Literature cited

Sanderson, M. J. (1997). A nonparametric approach to estimating divergence times in the absence of rate constancy. Molecular biology and evolution, 14(12), 1218-1231.

Sanderson, M. J. (2003). r8s: inferring absolute rates of molecular evolution and divergence times in the absence of a molecular clock. Bioinformatics, 19(2), 301-302.