bayroot: Bayesian sampling of HIV-1 integration dates by root-to-tip regression.

Area of Science:

• Virology
• Computational Biology
• Phylogenetics

Background:

• The latent human immunodeficiency virus 1 (HIV-1) reservoir composition is determined by proviral integration timing.
• Phylogenetic methods, such as root-to-tip (RTT) regression, can estimate these integration dates.
• Conventional RTT methods have limitations in accounting for molecular clock variation, root position uncertainty, and mutation rate heterogeneity.

Purpose of the Study:

• To develop a more accurate method for estimating HIV-1 proviral integration dates.
• To introduce an R package, bayroot, as a Bayesian extension of RTT.
• To enable incorporation of prior information on infection and antiretroviral therapy start times.

Main Methods:

• Implemented a Bayesian extension of RTT using a Metropolis-Hastings algorithm within an R package (bayroot).
• Input an unrooted maximum likelihood tree to sample the posterior distribution of molecular clock rate, root location, and root time.
• Used rejection sampling to simulate integration dates for HIV proviral sequences.
• Validated bayroot using the treeswithintrees (twt) package to simulate time-scaled trees from infected T cells.

Main Results:

• The bayroot package offers a Bayesian approach to estimate HIV-1 integration dates.
• bayroot successfully incorporates prior knowledge regarding infection and treatment timelines.
• Simulations demonstrated that bayroot provides significantly more accurate integration date estimates compared to conventional RTT.

Conclusions:

• bayroot improves the accuracy of estimating HIV-1 proviral integration dates.
• This method enhances our ability to study the dynamics of the latent HIV-1 reservoir.
• The bayroot R package is a valuable tool for HIV-1 molecular epidemiology and reservoir research.