A Composite Likelihood Approach to Inferring Phylogenetic Networks from Genomic Data

• Activity Number: 60 - Invited E-Poster Session II
• Type: Invited
• Date/Time: Sunday, August 8, 2021 : 6:45 PM to 7:30 PM
• Sponsor: Biometrics Section
• Author(s): Laura Kubatko* and Jing Peng and Sungsik Kong
• Companies: The Ohio State University and The Ohio State University and The Ohio State University
• Keywords: multispecies coalescent; phylogeny; phylogenetic network; composite likelihood; quartet

Abstract

Technological advances have led to an abundance of DNA sequence data that can be used to infer the evolutionary history of a collection of species. The multispecies coalescent model is commonly used for this purpose, as it captures variability in both the underlying evolutionary histories of genes and in the process of nucleotide substitution along these gene histories. However, the likelihood under the model cannot be efficiently computed for more than four species at a time. As a result, inference in the traditional likelihood or Bayesian frameworks cannot be carried out for problems of realistic size. An alternative approach is to use a composite likelihood constructed from the likelihood of all four-taxon subsets of species. The composite likelihood approach provides a method for estimating parameters along a fixed phylogenetic network that leads to asymptotically normal and statistically consistent estimators while still being computationally efficient. We use this approach to develop a stochastic algorithm for estimating phylogenetic networks under the multispecies coalescent model. We demonstrate the performance of our method using both simulated and empirical data.