Mapping the drivers of within-host pathogen evolution using massive data sets

Area of Science:

• Evolutionary biology
• Genetics
• Computational biology

Background:

• Host genetic variation and drug treatments can drive pathogen evolution within a host.
• Identifying host influences on pathogen evolution is challenging due to confounding genetic structures and multiple testing.
• Existing genetic association studies may lack the power to detect subtle host-pathogen interactions.

Purpose of the Study:

• To develop a robust Bayesian approach for detecting host genetic influences on pathogen evolution.
• To improve the power and precision of identifying host-pathogen interactions, particularly in large pathogen genomic datasets.
• To control for confounding factors like population structure in host and pathogen genomes.

Main Methods:

• A novel Bayesian statistical framework was developed to model pathogen evolutionary processes (recombination, selection).
• The method leverages large pathogen diversity datasets to enhance statistical power and account for population stratification.
• Simulations and empirical analysis of drug-induced selection in HIV-1 were used for validation.

Main Results:

• The Bayesian approach successfully identified known host-pathogen associations in simulations and empirical data.
• The method demonstrated superior precision-recall performance compared to existing approaches for detecting host influences.
• A high-resolution map of human leukocyte antigen (HLA)-induced selection on the HIV-1 genome was generated, revealing novel epitope-allele associations.

Conclusions:

• The developed Bayesian method provides a powerful and accurate tool for dissecting host influences on pathogen evolution.
• This approach can effectively identify specific regions of pathogen genomes affected by host factors, such as immune selection.
• The findings offer new insights into host-pathogen co-evolution and can guide the development of targeted therapeutic strategies.