Ancestral population genomics using coalescence hidden Markov models
24/10/2012 - Room A2-33 at 12:00 - External Seminar
(Bioinformatics Research Center (BiRC). )
The coalescence process is a very powerful model of neutral evolution, but combining the coalescence process with recombination results in a state space explosion that makes it infeasible to use in full genome analyses. The so-called sequential Markov coalescence (SMC) alleviates this by assuming the Markov property of the process when scanning along a genome alignment. The SMC was originally developed as a fast way of simulating sequences for the coalescence process, but several groups have recently started using it as a framework for constructing inference methods, combining the hidden Markov model (HMM) formalism with the SMC. This combination, called CoalHMMs, has proven to be a very powerful approach to do full genome population genetics.
In my talk I will explain how we use CoalHMMs in our group to explore the population genetics of ancestral species, such as the human and chimpanzee common ancestor. In such analyses, all intra-species genomes will generally have coalesced before the time period of interest, so we are usually restricted to a sample size of one from each species. Traditional analysis methods will not work because of this, but using full genomes in the CoalHMM framework enables us to explore the really deep coalescence time between species.
I will give examples from recently published great ape genomes: the orangutan genome (Devin et al. 2011), the gorilla genome (Scally et al. 2012), and the bonobo genome (Prüfer et al. 2012). I will also talk about how we are currently using CoalHMMs to explore the speciation process in recently diverged great apes such as the eastern and western gorillas, the Bornean and Sumatran orangutans, and bonobos and chimpanzees, and the split between humans and chimpazees.