Phylogenetic inference under recombination using Bayesian stochastic topology selection.

Area of Science:

• Genomics
• Computational Biology
• Evolutionary Biology

Background:

• Phylogenetic analysis traditionally assumes a single evolutionary relationship across genomes.
• Recombination, a key driver of diversity, can violate this assumption and lead to inaccurate phylogenetic inferences.
• Existing methods struggle to account for localized phylogenetic structures arising from recombination.

Purpose of the Study:

• To develop a generalized hidden Markov model (HMM) for inferring phylogenetic relationships along multiple sequence alignments.
• To account for rate heterogeneity and detect changes in phylogeny due to recombination.
• To overcome limitations of existing methods, such as the 4-taxa data restriction.

Main Methods:

• Generalization of a hidden Markov model (HMM) to infer phylogenies across genomic regions.
• Incorporation of Markov chain Monte Carlo (MCMC) algorithms to sample unknown model structures (number and type of topologies).
• Analytical integration over changepoints and unknown branch lengths for robust inference.

Main Results:

• Demonstrated the approach on simulated data, a human immunodeficiency virus (HIV) recombinant strain, and laboratory mouse strain sequences.
• Successfully distinguished between phylogenetic variation caused by recombination and rate heterogeneity.
• Validated the method's ability to handle complex genomic data beyond simple 4-taxa scenarios.

Conclusions:

• The generalized HMM provides a powerful tool for accurate phylogenetic inference in the presence of recombination.
• This method enhances the study of genome evolution by correctly identifying evolutionary relationships across diverse datasets.
• The approach offers a significant advancement for analyzing complex genomic data, including viral and mammalian genomes.