SNaQ.jl: Improved scalability for level-1 phylogenetic network inference

Area of Science:

• Computational Biology
• Evolutionary Biology
• Bioinformatics

Background:

• Phylogenetic networks model complex evolutionary events (hybridization, horizontal gene transfer) often missed by traditional trees.
• Existing phylogenetic network inference methods face scalability limitations due to computational demands and vast network search spaces.
• Composite likelihood methods like SNaQ offer improved tractability but are still insufficient for large datasets.

Purpose of the Study:

• Introduce SNaQ.jl, a Julia package enhancing the scalability of composite likelihood phylogenetic network inference.
• Improve computational efficiency for inferring complex evolutionary histories.

Main Methods:

• Implemented SNaQ.jl as a standalone Julia package.
• Integrated new scalability features: parallel quartet likelihood calculations, weighted random quartet selection, and probabilistic network search.
• Utilized composite likelihood inference.

Main Results:

• SNaQ.jl (version 1.1) demonstrates significant improvements in computational efficiency.
• Average runtimes were reduced by up to 499% compared to previous implementations.
• Accuracy and functional parameters remained unchanged, indicating enhanced performance without compromising results.

Conclusions:

• SNaQ.jl offers a computationally efficient solution for phylogenetic network inference.
• The package provides a scalable tool for analyzing complex evolutionary scenarios using composite likelihood methods.
• This advancement addresses the limitations of existing methods for large-scale phylogenetic analyses.