CoLaML: Inferring latent evolutionary modes from heterogeneous gene content

Abstract

MotivationEstimating the history of gene content evolution provides insights into genome evolution on a macroevolutionary timescale. Previous models did not consider heterogeneity in evolutionary patterns among gene families across different periods and/or clades. ResultsWe introduce CoLaML (joint inference of gene COntent evolution and its LA-tent modes using Maximum Likelihood), which considers heterogeneity using a Markov-modulated Markov chain. This model assumes that internal states determine evolutionary patterns (i.e., latent evolutionary modes) and attributes heterogeneity to their switchover during the evolutionary timeline. We developed a practical algorithm for model inference and validated its performance through simulations. CoLaML outperformed previous models in fitting empirical datasets and estimated plausible evolutionary histories, capturing heterogeneity among clades and gene families without prior knowledge. AvailabilityCoLaML is freely available at https://github.com/mtnouchi/colaml. Contactyamanouchi@bs.s.u-tokyo.ac.jp