POUMM: An R-package for Bayesian Inference of Phylogenetic Heritability

Abstract The Phylogenetic Ornstein-Uhlenbeck Mixed Model (POUMM) allows to estimate the phylogenetic heritability of continuous traits, to test hypotheses of neutral evolution versus stabilizing selection, to quantify the strength of stabilizing selection, to estimate measurement error and to make predictions about the evolution of a phenotype and phenotypic variation in a population. Despite this variety of applications, currently, there are no R-packages supporting POUMM inference on large non-ultrametric phylogenetic trees. Large phylogenies of that kind are becoming increasingly available, predominantly in epidemiology, where transmission trees are inferred from pathogen sequences during epidemic outbreaks, but also in some macroevolutionary studies incorporating fossil and contemporary data. In this article, we propose the R-package POUMM, providing Bayesian inference of the model parameters on large phylogenetic trees. We describe a novel breadth-first pruning algorithm for fast likelihood calculation, enabling highly parallelizable likelihood calculation on multi-core systems and GPUs. We report simulation-based results proving the technical correctness and performance of the software.