Bayesian analysis of amino acid substitution models

Area of Science:

• Computational Biology
• Bioinformatics
• Evolutionary Biology

Background:

• Amino acid substitution models are crucial for phylogenetic analysis but present challenges due to large parameter spaces.
• Existing fixed amino acid models, while reducing parameters, may not be suitable for all sequence alignments.

Purpose of the Study:

• To develop and evaluate novel approaches for analyzing amino acid sequences that address limitations of fixed models.
• To offer more flexible and accurate modeling of amino acid substitutions in evolutionary studies.

Main Methods:

• Exploration of a general time reversible model with Dirichlet priors on exchangeability parameters.
• Investigation of prior distributions centered on fixed amino acid model rates.
• Consideration of mixture models and partitioning schemes with Dirichlet process priors.

Main Results:

• The proposed methods offer alternatives to fixed amino acid models, allowing for more nuanced parameter estimation.
• These approaches enhance the adaptability of substitution models to diverse biological datasets.

Conclusions:

• New flexible models and prior distributions improve the analysis of amino acid sequences.
• These advancements provide researchers with better tools for phylogenetic inference and evolutionary modeling.