Inferring demographic parameters in bacterial genomic data using Bayesian and hybrid phylogenetic methods

Area of Science:

• Genomics
• Computational Biology
• Evolutionary Biology

Background:

• Next-generation sequencing enables rapid generation of large-scale pathogen genome data.
• Bayesian phylogenetic methods offer robust inference of evolutionary parameters but are computationally intensive.
• Computational demands of Bayesian methods limit their application in large-scale outbreak investigations.

Purpose of the Study:

• To compare the performance of fully Bayesian and hybrid phylogenetic inference methods.
• To evaluate the utility of hybrid methods for large-scale pathogen genome data analysis.
• To assess the reliability of hybrid approaches in phylodynamic and evolutionary timescale inference.

Main Methods:

• Analysis of six whole-genome single nucleotide polymorphism (SNP) datasets from bacterial species and simulations.
• Comparison of parameter estimates between fully Bayesian and hybrid phylogenetic inference frameworks.
• Implementation of a rapid maximum likelihood-based date-randomisation test to assess temporal signal.

Main Results:

• Hybrid and Bayesian methods produced highly similar estimates for evolutionary and demographic parameters.
• Congruence between methods depends on strong temporal signal (clock-like evolution) in the data.
• A rapid maximum likelihood date-randomisation test showed comparable performance to Bayesian counterparts.

Conclusions:

• Hybrid phylogenetic approaches provide a computationally efficient and reliable alternative to fully Bayesian methods.
• Hybrid methods significantly reduce analysis time for large datasets, making them suitable for outbreak studies.
• These findings support the use of hybrid methods for robust phylodynamic inference in pathogen surveillance.