DecentTree: scalable Neighbour-Joining for the genomic era

Area of Science:

• Computational Biology
• Bioinformatics
• Phylogenetics

Background:

• Neighbour-Joining is a widely used phylogenetic inference method.
• Current implementations struggle with datasets exceeding 10,000 sequences.
• The need for scalable phylogenetic tools is growing due to increasing sequence data generation.

Purpose of the Study:

• To develop a highly optimized and parallelized implementation of Neighbour-Joining.
• To provide a software solution that addresses the scalability limitations of existing phylogenetic inference methods.
• To create a tool that can be easily integrated into existing bioinformatics workflows.

Main Methods:

• Developed DecentTree, a stand-alone application and header-only library.
• Implemented optimized and parallelized versions of Neighbour-Joining and its variants.
• Benchmarked DecentTree against existing software like BIONJ, Quicktree, FastME, and RapidNJ.

Main Results:

• DecentTree demonstrates comparable or superior performance to existing software, particularly for large alignments.
• DecentTree is up to 6-fold faster than the fastest existing Neighbour-Joining software (RapidNJ).
• Achieved significant speed improvements when analyzing large datasets, such as 64,000 SARS-CoV-2 genomes.

Conclusions:

• DecentTree provides a scalable and efficient solution for phylogenetic inference using Neighbour-Joining.
• The software is suitable for analyzing large-scale genomic datasets, including those from disease outbreaks.
• DecentTree is available as open-source software, facilitating its adoption and integration.