Bayesian Phylogenomic Dating

The development of divergence-time estimation methods has been an active area of research since the early 1960s, when the molecular clock was first postulated by Zuckerkandl and Pauling. Thanks to technological and computational improvements, more powerful and cutting-edge techniques and algorithms have been developed to better understand species evolution at the molecular level. These have led to improved methods for molecular clock dating of speciation events. During the past two decades, the approaches for DNA sequencing have substantially advanced and their costs have decreased, thus enabling large-scale genome-sequencing projects that aim to sequence all species in the tree of life. Being able to access thousands of complete genomes, however, has brought new biological and computational challenges to phylogenomic analyses. We might have more data, but also new questions to answer. Inferring reliable phylogenies and accurately dating them is now the main goal of phylogenomic analyses. Although new computational tools that implement more complex evolutionary models have been developed, there remain challenges in dealing with issues such as polytomies, incomplete lineage sorting, and the uncertainty in the fossil record. This chapter aims to guide the reader through the steps of Bayesian phylogenomic dating analyses, from data collection and processing up to the inference of the species tree and subsequent clock dating analysis. We pay close attention to the Bayesian paradigm in molecular clock dating, focusing on the effects that the prior and the likelihood can have on the estimated divergence times when using phylogenomic data. We describe strategies to speed up computation when using large genomic data sets, such as the approximate-likelihood method, which produces speed-ups of up to 1000× in time-tree inference. We also discuss strategies to improve the efficiency of Markov chain Monte Carlo sampling.