Making the Most of Data Augmentation in Phylogenetics

• Activity Number: 292
• Type: Invited
• Date/Time: Tuesday, August 11, 2015 : 8:30 AM to 10:20 AM
• Sponsor: International Indian Statistical Association
• Author(s): Vladimir Minin*
• Companies: University of Washington
• Keywords: stochastic mapping ; uniformization ; data imputation ; markov ; missing data ; molecular evolution

Abstract

Phylogenetic inference, concerned with reconstructing evolutionary relationships among species/organisms, often involves summing/integrating over unobserved evolutionary paths undertaken by the traits of interest. Augmenting observed data with these hidden paths allowed phylogeneticists to take advantage of established computational statistics tools, such as expectation-maximization and Markov chain Monte Carlo (MCMC) data augmentation algorithms. We review these developments and highlight our recent contribution to this algorithmic line of work. However, data augmentation has not only algorithmic, but also statistical advantages. For examples, we will see that data augmentation can help estimating evolutionary rate parameters under mild misspecification of a phylogenetic model. Moreover, we will consider a strategy that uses data augmentation to construct new phylogenetic estimation procedures that are more flexible that likelihood-based competitors.