PLACE:  W142 Lagomarcino

SPEAKER:
Karin Dorman
University of California

TITLE:
Is in vivo recombination between HIV strains ovrestimated?

ABSTRACT:

Recombination between HIV viral genomes is invoked to explain how sequences from the same virus can shift position in phylogenetic trees inferred from alternate genes or regions.  Since the first HIV recombinant was discovered in 1988, improved detection methods have generated evidence that over 10% of sequences in the HIV database are recombinant. Incorrect topological inference, however, can lead to false evidence of recombination and possible overestimation of the recombination rate.  Topological inference error rates
increase when statistical variation is exacerbated by short sequence segments between recombination points or evolutionary conditions that violate the assumed models of evolution.  We use simulations to demonstrate that sequence lengths, levels of divergence, and evolutionary assumption violations consistent with HIV evolution and recombination lead to substantial error rates for three common phylogenetic methods and confidence levels used in HIV research. In order to estimate a p-value for recombination, we adapt a
bootstrap method for regions to assess sequence data with hypothesized recombinant breakpoints.  We also estimate the odds and posterior probability for recombination, two quick and computationally simpler methods for assessing confidence.  We analyze 73 putative recombinants using these methods and find
only 71% are supported at a significance level of 0.05.
 
 
 

COFFEE: 3:45 p.m., MBB Atrium