paPAML

paPAML analyzes selection pressures on protein-coding sequences by extending the Phylogenetic Analysis by Maximum Likelihood (PAML) codeml models to compile and accelerate comparative analyses of neutral and selected DNA changes.

Key Features:

• CDS Extraction: Includes a dedicated CDS extractor module to compile protein-coding sequences for downstream analysis.
• Parallel Processing: Executes analyses across multiple computing threads to accelerate model runs.
• Comprehensive Model Analysis: Compiles results from site-specific, branch-site, and branch models to examine selection across different evolutionary scenarios.
• Detection of Negative Selection: Identifies sites under negative selection and outputs codon lists with significance values.
• PAML Integration: Builds upon the Phylogenetic Analysis by Maximum Likelihood (PAML) framework, invoking codeml model implementations.
• Performance and Speed: When tested with 10 computing threads, produced identical selection results while completing tasks approximately 3–5× faster than the longest-running codeml model and 7–15× faster than HyPhy FEL.

Scientific Applications:

• Evolutionary selection analysis: Comparative analysis of neutral and selected changes in protein-coding DNA sequences across phylogenies.
• Episodic constrained, relaxed, and adaptive change detection: Detection and characterization of episodic shifts in selective regimes on coding sequences.
• GINS Complex Subunit 3 exon studies: Applied to investigate selection patterns in the GINS Complex Subunit 3 exon.
• Neutrophil-associated gene analysis: Applied to assess selection pressures on neutrophil-associated genes.
• Lysin gene analysis: Applied to evaluate selection dynamics in lysin.
• Apolipoprotein gene analysis: Applied to investigate selection on apolipoprotein genes.

Methodology:

Uses a CDS extractor, runs PAML codeml models (site-specific, branch-site, branch) in parallel across multiple threads, compiles model results and outputs codon lists with significance values, and was benchmarked against codeml and HyPhy FEL with reported 10-thread speedups of ~3–5× versus the longest-running codeml model and ~7–15× versus HyPhy FEL.