Maximum likelihood pandemic-scale phylogenetics.

Area of Science:

• Genomic epidemiology
• Computational biology
• Virology

Background:

• Phylogenetic analyses are vital for understanding viral evolution and spread, particularly during pandemics such as COVID-19.
• Existing phylogenetic methods struggle to handle the massive genomic datasets generated during large-scale outbreaks.
• The accurate inference of viral origins, transmission dynamics, and variant emergence is hindered by computational limitations.

Purpose of the Study:

• To introduce a novel computational approach, MAximum Parsimonious Likelihood Estimation (MAPLE), for scalable phylogenetic analysis.
• To enable accurate and efficient phylogenetic inference on unprecedented scales of genomic data.
• To advance the application of phylogenetics in genomic epidemiology for large-scale infectious disease surveillance.

Main Methods:

• Development of MAximum Parsimonious Likelihood Estimation (MAPLE), a likelihood-based phylogenetic inference method.
• Comparative analysis of MAPLE against existing maximum likelihood phylogenetic approaches using large SARS-CoV-2 genomic datasets.
• Evaluation of computational performance in terms of speed and memory usage.

Main Results:

• MAPLE demonstrates superior accuracy in inferring SARS-CoV-2 phylogenies compared to current maximum likelihood methods.
• MAPLE achieves significant speed improvements, running up to thousands of times faster than existing approaches.
• MAPLE requires substantially less memory, at least 100 times less, for analyzing large genomic datasets.

Conclusions:

• MAPLE offers a scalable and accurate solution for phylogenetic analysis of massive genomic datasets.
• The method significantly enhances the capabilities of genomic epidemiology, facilitating the analysis of millions of pathogen genomes.
• MAPLE will enable continued robust phylogenetic, phylogeographic, and phylodynamic studies crucial for pandemic response and preparedness.