modelgenerator

modelgenerator selects amino acid and nucleotide substitution models for maximum likelihood phylogenetic tree inference by evaluating dataset-specific patterns of substitution.

Key Features:

• Empirical Model Evaluation: Derives the maximum likelihood substitution model directly from the input dataset for amino acid and nucleotide sequences.
• Robust Statistical Methods: Implements statistical model-selection methods validated in simulations to evaluate and compare substitution models.
• Dataset-Specific Optimization: Assesses patterns of amino acid substitution among homologous sequences to identify the most suitable substitution matrix for each dataset.
• Cross-Domain Applicability: Identifies models that can apply across domains, exemplified by a retroviral Pol-derived model fitting large proteobacterial and archaeal datasets.

Scientific Applications:

• Phylogenetic tree reconstruction: Supports maximum likelihood phylogenetic analyses by selecting appropriate substitution models for amino acid and nucleotide data.
• Comparative evolutionary analyses: Facilitates evolutionary inference across proteobacteria, archaea, and viral proteins through empirical model selection.
• Analysis of complex or heterogeneous datasets: Improves reliability of evolutionary conclusions by tailoring substitution models to dataset-specific substitution patterns.

Methodology:

Analyzing empirical patterns of amino acid substitution within the dataset.
Utilizing statistical methods to evaluate and compare different substitution models.
Highlighting potential pitfalls of arbitrary model selection through empirical examples.