bayroot

bayroot estimates integration dates of human immunodeficiency virus 1 (HIV-1) proviruses into host genomes using a Bayesian framework to infer temporal dynamics of the latent reservoir.

Key Features:

• Bayesian Framework: Extends root-to-tip (RTT) regression by implementing a Bayesian framework that incorporates prior information on infection timelines and antiretroviral therapy (ART) initiation.
• Handling Uncertainties: Accommodates variation in mutation rates over time, accounts for uncertainties in molecular clock estimation, and allows flexible determination of root position in phylogenetic trees.
• Metropolis-Hastings Algorithm: Implements a Metropolis-Hastings algorithm to sample the joint posterior distribution of the molecular clock rate, tree root location, and root time.
• Rejection Sampling: Applies rejection sampling to simulate integration dates for proviral sequences from the posterior distribution of model parameters.
• Implementation: Provided as an R package.

Scientific Applications:

• Integration Date Estimation: Provides improved proviral integration date estimates compared to conventional RTT methods.
• Latent Reservoir Analysis: Enables characterization of the composition and temporal evolution of the HIV-1 latent reservoir within a host.
• Incorporation of Clinical Timelines: Allows incorporation of infection timelines and ART initiation dates to constrain temporal estimates.
• Validation with Simulations: Validated using simulated time-scaled trees that relate samples from actively and latently infected T cells within a single host.

Methodology:

Takes an unrooted maximum likelihood tree as input, uses Bayesian inference with priors on infection timelines and ART initiation, employs a Metropolis-Hastings sampler for the joint posterior of molecular clock rate, root location, and root time, and then uses rejection sampling to simulate integration dates from the posterior.