MODELESTIMATOR

MODELESTIMATOR estimates amino acid replacement rate matrices from aligned protein sequences to model protein evolutionary processes.

Key Features:

• Data-Specific Adaptation: Generates data-specific amino acid replacement rate matrices for positions, protein families, lineages, or protein regions where general matrices may be inadequate.
• Pairwise-based Estimation Methods: Implements two novel methods to estimate replacement rate matrices from independent pairwise protein sequence alignments and outperforms Müller–Vingron's resolvent method in tests.
• Robustness to Dataset Size and Divergence: Operates on small to very large datasets and uses all sequence pairs regardless of length or divergence to maximize data utilization.
• Generalization to Multialignments: Extends to multialignment data to derive family-specific rate matrices that frequently yield higher likelihoods than general matrices.

Scientific Applications:

• Phylogenetic Analysis: Provides tailored replacement matrices to improve evolutionary model fit for phylogenetic tree inference.
• Protein Family Studies: Derives family-specific rate matrices to examine evolutionary dynamics and constraints within protein families.
• Comparative Genomics: Accommodates lineage- or region-specific evolutionary rates to facilitate comparisons across lineages in the tree of life.

Methodology:

Estimates replacement rate matrices from independent pairwise protein sequence alignments using two novel estimation methods, compares performance to Müller–Vingron's resolvent method, validates results on synthetic datasets, and can be generalized to multialignment data.