Inferring heterogeneous evolutionary processes through time: from sequence substitution to phylogeography.

Area of Science:

• Evolutionary biology
• Computational phylogenetics
• Molecular evolution

Background:

• Phylogenetic and phylogeographic analyses typically assume constant evolutionary rates over time.
• This time-homogeneity assumption simplifies computation but limits biological realism and the study of evolutionary dynamics.

Purpose of the Study:

• To develop and implement an evolutionary model that relaxes the time-homogeneity assumption.
• To allow for different substitution rate matrices across distinct time intervals (epochs).
• To infer temporal dynamics in evolutionary histories for discrete data types.

Main Methods:

• Developed an epoch model within a Bayesian inference framework.
• Implemented a massively parallel approach using graphics processing units (GPUs) for computational efficiency.
• Assessed model performance using synthetic nucleotide and codon substitution data.

Main Results:

• The epoch model successfully recovers evolutionary parameters from simulated data under various evolutionary scenarios.
• Demonstrated the model's ability to capture key features of temporal heterogeneity in empirical data.
• Validated the model's performance in analyzing HIV within-host evolution and influenza seasonality.

Conclusions:

• The proposed epoch model offers a more biologically realistic approach to phylogenetic and phylogeographic inference.
• The massively parallel implementation makes complex temporal heterogeneity analysis computationally feasible.
• This framework provides a powerful tool for investigating dynamic evolutionary processes.