DISCO: Species Tree Inference using Multicopy Gene Family Tree Decomposition.

Area of Science:

• Computational Biology
• Phylogenetics
• Bioinformatics

Background:

• Species tree inference is complex due to gene tree heterogeneity caused by gene duplication and loss.
• Existing methods for gene duplication and loss are often computationally expensive.

Purpose of the Study:

• To introduce DISCO, a new method for decomposing gene family trees to facilitate species tree inference.
• To evaluate the accuracy and efficiency of DISCO when combined with various species tree estimation methods.

Main Methods:

• DISCO decomposes multi-copy gene family trees into single-copy trees.
• Species tree inference was performed using DISCO with methods like ASTRAL, ASTRID, and IQ-TREE.
• Statistical consistency of ASTRAL-DISCO was proven under the gene duplication and loss (GDL) model.

Main Results:

• ASTRAL-DISCO demonstrates statistical consistency under the GDL model.
• DISCO paired with different methods achieves high accuracy across various model conditions.
• DISCO outperforms the MI decomposition strategy in accuracy.

Conclusions:

• DISCO provides an accurate and computationally efficient approach for species tree inference.
• The method effectively handles gene tree heterogeneity arising from gene duplication and loss.
• DISCO enhances the applicability of existing single-copy gene tree methods to complex phylogenetic problems.