bModelTest

bModelTest infers and compares Bayesian phylogenetic site models for nucleotide sequence data to enable joint inference and marginalization of substitution models during MCMC.

Key Features:

• Bayesian Framework: Implements Bayesian inference to integrate uncertainty and compute posterior probabilities for site model parameters.
• Trans-Dimensional MCMC Proposals: Employs trans-dimensional Markov chain Monte Carlo proposals to switch dynamically between substitution models during analysis.
• Comprehensive Model Inference: Estimates posterior probabilities for gamma-distributed rate heterogeneity, proportions of invariant sites, and unequal base frequencies.
• Flexibility with Substitution Models: Supports the full set of time-reversible nucleotide substitution models and two specified subsets for targeted applications.

Scientific Applications:

• Phylogenetic model selection: Enables inference and marginalization over site models during MCMC to improve model selection for nucleotide sequence analyses.
• Phylogenetic reconstruction: Facilitates more precise and reliable tree reconstruction by jointly inferring site models and phylogenies under diverse substitution models.

Methodology:

bModelTest integrates with BEAST 2 and uses a Bayesian framework with trans-dimensional MCMC proposals to perform joint inference of site models and phylogenetic trees across time-reversible nucleotide models.