General Departmental Seminar Series
Identification of Recombinant HIV Sequences Using a Bayesian Multiple Break Point Phylogenetic Model
Rob Weiss, PhD, Department of Biostatistics, UCLA School of Public Health
Based on work by: MA Suchard, RE Weiss, KS Dorman, JS Sinsheimer
Friday, Oct 25, 2002, 12:00 pm
G5/142 Clinical Sciences Center, 600 Highland Ave.
We develop methods for identifying recombination and simultaneously inferring parentals and crossover points (COPs) using a Bayesian multiple change-point model. The model assumes that the sites along the data separate into an unknown number of contiguous segments, where segments may support alternative topologies, average rates of evolution and nucleotide transition-transversion ratios. The model also returns measures of uncertainty in all model parameters, including COPs; whereas current methods yield at best point estimates. We employ reversible jump Markov chain Monte Carlo to fit our model and test for recombination using Bayes factors.
We illustrate the methodology on several HIV-1 examples. We first examine the entire genome from an HIV-1 isolate previously identified as a recombinant and the causative agent of an epidemic outbreak of HIV-1 infection among intravenous drug users. We find strong evidence for multiple crossovers along the genome, with an inferred parental heritage of ABABA, and a high level of selective pressure changes within the vif through env genes. We then simultaneously fit our model and test for recombination in three putative gag gene recombinants. HIV-1 isolate RW024 decisively supports recombination with an inferred parental heritage of AD. HIV-1 isolate VI557 barely supports recombination.
Back to General Departmental Seminar Series