Phylogeny Estimation by Integration over Isolation with Migration Models

Abstract

Phylogeny estimation is difficult for closely related populations and species, especially if they have been exchanging genes. We present a hierarchical Bayesian, Markov-chain Monte Carlo method with a state space that includes all possible phylogenies in a full Isolation-with-Migration model framework. The method is based on a new type of genealogy augmentation called a “hidden genealogy” that enables efficient updating of the phylogeny. This is the first likelihood-based method to fully incorporate directional gene flow and genetic drift for estimation of a species or population phylogeny. Application to human hunter-gatherer populations from Africa revealed a clear phylogenetic history, with strong support for gene exchange with an unsampled ghost population, and relatively ancient divergence between a ghost population and modern human populations, consistent with human/archaic divergence. In contrast, a study of five chimpanzee populations reveals a clear phylogeny with several pairs of populations having exchanged DNA, but does not support a history with an unsampled ghost population.

Models of archaic admixture and recent history from two-locus statistics

“…Modeling archaic admixture worldwide resolved this discrepancy. We recovered the signal of Neanderthal admixture in Eurasian populations, and found evidence for substantial and long-lasting admixture from a deeply diverged lineage in two African populations that is consistent with evidence from previous studies (Plagnol and Wall, 2006;Skoglund et al, 2017;Hey et al, 2018;Durvasula and Sankararaman, 2018).…”

“…could also contribute to Eurasians through admixture prior to the OOA event (Figure 4(A)). Many human lineages coexisted on the African continent, possibly until quite recently (Rightmire, 2009;Harvati et al, 2011;Berger et al, 2017), and genetic evidence points to a history of archaic admixture or deep structure across many modern African populations (Hammer et al, 2011;Lachance et al, 2012;Hsieh et al, 2016;Skoglund et al, 2017;Durvasula and Sankararaman, 2018;Hey et al, 2018). It is likely that modern humans have met and mixed with diverged lineages many times through history, rather than receiving just a single pulse of migrants (Browning et al, 2018;Villanea and Schraiber, 2019).…”

A Bayesian implementation of the multispecies coalescent model with introgression for comparative genomic analysis

“…Sites within the same locus are assumed to share the same genealogical history, while the gene trees and coalescent times are assumed to be independent among loci given the species tree and parameters. The ideal data for this kind of analysis are loosely linked short genomic segments (called loci), so that recombination within a locus is unimportant while different loci are largely independent (Burgess and Yang, 2008;Hey et al, 2018;Lohse et al, 2016). The Bayesian formulation consists of two components: (i) the probability density of gene trees given the species tree under the MSci model, f (G i |τ τ τ, θ θ θ , ϕ ϕ ϕ), given in Yu et al (2014): Note that this density differs from that given by Kubatko (2009), as pointed out by Solis-Lemus and Ane (2016); and (ii) the likelihood of the sequence data at each locus i given the gene tree, f (X i |G i ) (Felsenstein, 1981).…”

“…Full-likelihood and summary methods to accommodate introgression/migration While biologically simplistic, the MSci and IM models offer powerful tools for analysis of genomic sequence data from closely related species, when cross-species gene flow appears to be the norm (Mallet et al, 2016;Martin and Jiggins, 2017). Full likelihood implementations of those models, including the ML (Dalquen et al, 2017;Zhu and Yang, 2012) and Bayesian MCMC methods (Hey et al, 2018;Wen and Nakhleh, 2018;Zhang et al, 2018), make efficient use of the information in the data and naturally accommodate phylogenetic uncertainties at individual loci caused by high sequence similarities (Edwards et al, 2016;Xu and Yang, 2016). The complexity of those models means that large datasets with hundreds or thousands of loci may be necessary to obtain reliable parameter estimates, as indicated by our analyses of both simulated and real data.…”

“…The space of unknown genealogical histories (including the migration/introgression histories) is large, and constraints between the species tree and the gene trees make it difficult to traverse the parameter space in the posterior. Current MCMC implementations include IMa3 (Hey, 2010;Hey et al, 2018) c The Author 2019. for the IM model, and *BEAST (Jones, 2019;Zhang et al, 2018) and PHYLONET (Wen and Nakhleh, 2018) for the MSci model. It does not appear computationally feasible to apply those programs to realistically sized datasets, with more than 200 loci, say.…”