Coalescent Inference Using Serially Sampled, High-Throughput Sequencing Data from Intrahost HIV Infection

Area of Science:

• Population Genetics
• Virology
• Computational Biology

Background:

• Human immunodeficiency virus (HIV) exhibits rapid evolution and high intrahost genetic diversity.
• Deep sequencing provides detailed insights into viral evolution over time.
• Inferring population genomics from HIV sequence data is complex due to high mutation/recombination rates and demographic changes.

Purpose of the Study:

• To develop a new computational method for inferring evolutionary dynamics from HIV deep sequencing data.
• To accurately estimate mutation rates, recombination rates, and effective population size within HIV infections.
• To handle challenges such as sampling across multiple time points and missing data.

Main Methods:

• Developed a novel inference method using importance sampling of ancestral recombination graphs under a multilocus coalescent model.
• Extended approximations for conditional sampling distributions to improve coalescent likelihood calculations.
• Implemented the method in freely available software called Coalescenator.

Main Results:

• The Coalescenator method successfully inferred mutation rates and effective population size, yielding results comparable to existing software (BEAST).
• The method demonstrated the ability to estimate local recombination rates.
• Successfully applied to HIV-1 deep sequencing data from an infected individual sampled over seven time points across 2 years.

Conclusions:

• Coalescenator offers a powerful and computationally efficient tool for analyzing complex HIV population genomic data.
• The method advances the inference of intrahost viral evolutionary dynamics, including mutation, recombination, and population size.
• Provides a valuable resource for researchers studying HIV evolution and developing therapeutic strategies.