MCMCcodonsite

MCMCcodonsite implements codon-based substitution models within a Bayesian framework to estimate site-specific nonsynonymous/synonymous rate ratios (ω = dN/dS) and detect sites under positive selection.

Key Features:

• Codon models: Supports codon substitution models M0, M2a, M3, M7, and M8 for estimating ω across sites.
• Bayesian framework: Fits probability distributions that allow some sites to have ω > 1 to identify positively selected sites while controlling false-positive rates.
• Empirical Bayes and full-Bayes: Implements both empirical Bayes and full-Bayes approaches, with full-Bayes providing improved performance on small or simulated datasets.
• Heuristic acceleration: Employs heuristics to streamline computations for large datasets without compromising reported accuracy.
• Model sensitivity and selection: Notes sensitivity to model specification and supports model selection guidance using the Akaike Information Criterion (AIC).

Scientific Applications:

• Detection of positive selection: Identifies sites with ω > 1 in protein-coding genes to infer adaptive evolution.
• Site-specific ω estimation: Estimates sitewise nonsynonymous/synonymous rate ratios (ω = dN/dS) across gene alignments.
• Evolutionary and comparative genomics: Applies to evolutionary biology and genomics studies for inferring selective pressures and adaptive changes.
• Analysis of small or simulated datasets: Uses full-Bayes methods to improve inference accuracy on small sample sizes or simulated data.

Methodology:

Implements Bayesian fitting of probability distributions that permit ω > 1, evaluates codon models M0, M2a, M3, M7, and M8, provides empirical Bayes and full-Bayes inference, incorporates computational heuristics, and supports model selection via the Akaike Information Criterion (AIC).