Extensively Parameterized Mutation-Selection Models Reliably Capture Site-Specific Selective Constraint.

Area of Science:

• Evolutionary biology
• Computational biology
• Genomics

Background:

• Mutation-selection models are key for understanding coding sequence evolution.
• Two frameworks exist: fixed-effects maximum likelihood and random-effects Bayesian Dirichlet Process.
• These frameworks show differing predictions for selection coefficient distributions.

Purpose of the Study:

• To evaluate the accuracy of fixed-effects and random-effects models in inferring site-specific evolutionary constraints.
• To compare the performance of these two mutation-selection inference frameworks.

Main Methods:

• A simulation-based strategy was employed to assess accuracy.
• The study compared how well each approach recapitulated site-specific selective constraints.
• Inferences from both frameworks were analyzed for site-specific selective constraint.

Main Results:

• The fixed-effects approach accurately estimates site-specific evolutionary constraint.
• The random-effects Bayesian approach systematically underestimates natural selection, especially at slowly evolving sites.
• Despite differing selection coefficient distributions, both frameworks produced similar site-specific constraint inferences.

Conclusions:

• The fixed-effects mutation-selection framework is more reliable for evolutionary analysis.
• This framework is recommended for future applications and development in estimating evolutionary constraints.
• Accurate site-specific constraint inference is crucial for understanding evolutionary processes.