Mutation-selection models of coding sequence evolution with site-heterogeneous amino acid fitness profiles

Area of Science:

• Evolutionary biology
• Molecular evolution
• Computational biology

Background:

• Codon-based evolutionary models are crucial for studying long-range evolutionary patterns.
• Existing models fail to fully integrate site-specific selection effects observed at the amino acid level.
• Current codon-level studies are either too simplistic or over-parameterized.

Purpose of the Study:

• To propose a novel probabilistic model that captures the heterogeneity of amino acid fitness profiles across gene coding positions.
• To integrate recent nonparametric statistical developments into evolutionary modeling.
• To provide a more biologically plausible framework for evolutionary studies.

Main Methods:

• Developed a probabilistic model incorporating nonparametric statistical approaches.
• Accounted for heterogeneity of amino acid fitness profiles across coding positions.
• Applied the model to multiple real protein-coding gene alignments.

Main Results:

• The model produces biologically plausible inferences regarding site-specific amino acid constraints.
• It accurately estimates distributions of scaled selection coefficients.
• Demonstrated the model's effectiveness on diverse protein-coding gene alignments.

Conclusions:

• The proposed model successfully accounts for heterogeneous selection regimes at the amino acid level.
• This framework can be extended to include other evolutionary factors like mutation biases and population size.
• Offers a robust foundation for future advancements in population-genetic modeling.