BAli-Phy

BAli-Phy performs Bayesian joint estimation of molecular sequence alignments and phylogenetic trees using Markov chain Monte Carlo to account for alignment uncertainty and indel processes.

Key Features:

• Simultaneous Estimation: Considers the joint space of alignments and phylogenies rather than relying on a single alignment, eliminating the need for guide trees and reducing bias from progressive alignment methods.
• Incorporation of Indel Information: Integrates insertion/deletion (indel) data, using affine gap penalties, to improve phylogenetic resolution and grouping of sister taxa, particularly in rapidly evolving pathogens.
• Modeling Rate Variation: Models rate variation between sites during alignment and phylogeny estimation.
• Novel MCMC Transition Kernel: Uses a transition kernel that enhances MCMC mixing, supports non-local topology rearrangements, and enables block sampling of alignment and topology parameters to facilitate convergence from arbitrary starting alignments.
• Alignment Uncertainty Consideration: Estimates alignment uncertainty and provides methods to summarize that uncertainty.
• Codon-based Substitution Models: Addresses biases introduced by codon-based substitution models that can constrain indels to occur between codons.
• Bayesian Framework for Positive Selection Analysis: Jointly estimates the degree of positive selection and sequence alignment under an evolutionary model to reduce false positives in detecting diversifying selection.
• Inference Using Bayes Factors: Employs Bayes factors for model comparison across differing dimensionality and uses Rao-Blackwellization for more accurate estimation.

Scientific Applications:

• Phylogenetic analysis of rapidly evolving pathogens: Enhances phylogenetic resolution by leveraging indel information and accounting for alignment uncertainty in datasets with limited substitution signal.
• Positive selection detection: Reduces false positives in diversifying positive selection analyses by integrating over alignment uncertainty.
• Evolutionary biology research: Provides joint alignment and phylogeny estimation under evolutionary models for studies requiring integrated treatment of substitutions and indels.

Methodology:

Markov chain Monte Carlo exploration of the joint alignment–phylogeny space; indel modeling with affine gap penalties; modeling of rate variation between sites; a novel MCMC transition kernel enabling non-local topology rearrangements and block sampling; use of codon-based substitution models; Bayesian model comparison with Bayes factors and Rao-Blackwellization; joint estimation of selection parameters and alignment.