PCMBase

PCMBase implements efficient likelihood calculations and simulations of multivariate Gaussian phylogenetic models to study the evolution of quantitative traits across phylogenies.

Key Features:

• Efficient likelihood calculation: Implements an algorithm whose computational cost is linear in the number of nodes of a phylogenetic tree for multivariate Gaussian models.
• Handling heterogeneous evolutionary forces: Supports parameter "shifts" at specified tree locations, allowing changes to any parameters or transitions between models within the G_LInv family.
• Support for complex tree structures: Accommodates polytomies and non-ultrametric trees.
• Robustness to missing data: Handles datasets with missing trait values during likelihood calculation and simulation.
• Integration with inference tools: Provides a generic library interface for use with maximum likelihood and Bayesian inference methods.

Scientific Applications:

• Evolution of quantitative traits: Enables comparative analyses of quantitative trait evolution across organisms, including microorganisms, plants, and animals.
• Modeling heterogeneous evolutionary processes: Facilitates detection and modeling of shifts in evolutionary parameters across a phylogeny.
• Large-scale phylogenetic analyses: Supports likelihood-based inference on large trees spanning hundreds to thousands of species.

Methodology:

Based on the G_LInv sub-family of Gaussian phylogenetic models, PCMBase uses transition densities with expectations linear in the ancestral trait value and variances invariant to that value, and implements a likelihood-calculation algorithm with complexity linear in the number of tree nodes while supporting parameter and model shifts within G_LInv.