Weighted neighbor joining: a likelihood-based approach to distance-based phylogeny reconstruction

Area of Science:

• Computational Biology
• Evolutionary Biology
• Bioinformatics

Background:

• Phylogeny reconstruction is crucial for understanding evolutionary relationships.
• Existing methods like neighbor joining can be susceptible to errors with long evolutionary distances.
• Maximum-likelihood methods are accurate but computationally intensive.

Purpose of the Study:

• Introduce a new distance-based phylogeny reconstruction method, Weighbor (weighted neighbor joining).
• Address limitations of traditional neighbor joining, particularly concerning distance estimation errors.
• Develop a faster yet accurate alternative to maximum-likelihood methods.

Main Methods:

• Developed a weighted neighbor joining criterion accounting for exponentially larger errors in longer distance estimates.
• Modeled distances as correlated Gaussian random variables under a probabilistic model of sequence evolution.
• Incorporated an additivity term and a positivity term into the Weighbor criterion to evaluate implications of joining taxa.

Main Results:

• Weighbor reconstructs phylogenies significantly faster than maximum-likelihood methods.
• The trees generated by Weighbor are qualitatively and quantitatively similar to those from maximum-likelihood analysis.
• Weighbor demonstrates robustness against the 'long branches attract' and 'long branch distracts' problems.

Conclusions:

• Weighbor offers a computationally efficient and accurate approach to distance-based phylogeny reconstruction.
• The method effectively mitigates common artifacts in phylogenetic tree building.
• Weighbor provides a promising alternative for large-scale evolutionary analyses.