mrmodeltest2

mrmodeltest2 applies Bayesian Markov chain Monte Carlo (MCMC) to select and compare DNA substitution models and evaluate process heterogeneity for phylogenetic inference.

Key Features:

• Bayesian MCMC approach: Employs a Bayesian MCMC framework to explore and compare substitution models on combined datasets.
• Model complexity and partitioning: Supports partitioning across sites, independent parameters for each gene, and model complexities spanning morphological and molecular data.
• Among-site rate variation and process heterogeneity: Evaluates among-site rate variation and heterogeneity of evolutionary processes across data partitions.
• Convergence and mixing: Maintains convergence and parameter mixing across complex, parameter-rich models.
• Incorporation of morphological data: Integrates morphological characters (noting that ~5% of characters can substantially affect combined-data trees) alongside genetic data.
• Parameter update cycle: Uses an efficient parameter update cycle robust to model partitioning across sites.

Scientific Applications:

• Phylogenetic inference: Infers phylogenetic relationships using combined morphological and genetic data, including nuclear and mitochondrial genes and ribosomal and protein-coding sequences.
• Model selection with Bayes factors: Uses Bayes factors to evaluate model complexity and to assess the significance of process heterogeneity and among-site rate variation.
• Balancing complexity and accuracy: Assesses trade-offs between model complexity, topological uncertainty, and parameter estimation accuracy to identify appropriate substitution models.

Methodology:

Performs Bayesian MCMC on combined datasets with partitioning and independent per-gene parameters, integrates morphological and genetic data, uses Bayes factors to compare model complexity and among-site rate variation, and employs an efficient parameter update cycle to promote convergence and mixing.