Belvu

Belvu reconstructs phylogenetic trees from multiple sequence alignments using the Scoredist protein distance estimator to provide accurate and robust evolutionary distance estimates.

Key Features:

• Scoredist Distance Estimator: Developed as a correction-based method that applies a logarithmic correction to observed divergence using the BLOSUM62 score matrix and accounts for multiple substitutions within alignments.
• Model Independence: Estimates distances without relying on transition probability matrices (e.g., Dayhoff series), yielding consistent results across evolutionary models.
• Performance and Robustness: Evaluated against optimal matrix methods with training and testing across Dayhoff, Jones-Taylor-Thornton, and Müller-Vingron models and testing on Whelan and Goldman, showing accuracy comparable to optimal matrix methods and superior robustness when trained on one model but tested on another.
• Efficiency: Fast to implement and execute, providing a computationally efficient alternative to correction-based methods such as Jukes-Cantor and Kimura.

Scientific Applications:

• Comparative Genomics: Analyzing evolutionary relationships among species or genes using phylogenetic trees derived from multiple sequence alignments.
• Molecular Evolution Studies: Investigating patterns of sequence divergence and adaptation via estimated evolutionary distances.
• Protein Structure Prediction: Inferring structural similarities based on evolutionary distance estimates.

Methodology:

Scoredist uses the BLOSUM62 matrix to apply a logarithmic correction to observed divergences to estimate true substitutions in aligned protein sequences; it was evaluated by training and testing on Dayhoff, Jones-Taylor-Thornton, and Müller-Vingron models and tested on Whelan and Goldman and compared to optimal matrix methods.