|
Activity Number:
|
15
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Sunday, August 8, 2004 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #301050 |
|
Title:
|
Bayesian Recombination Identification: New Models and Better Ways of Incorporating Prior Information
|
|
Author(s):
|
Vladimir N. Minin*+ and Karin S. Dorman and Marc A. Suchard
|
|
Companies:
|
University of California, Los Angeles and Iowa State University and University of California, Los Angeles
|
|
Address:
|
Department of Biomathematics , Los Angeles, CA, 90095,
|
|
Keywords:
|
recombination ; HIV evolution ; Bayes ; MCMC ; spatial point process
|
|
Abstract:
|
We develop a dual Multiple Change Point (MCP) model in a Bayesian framework to improve detection of recombination among aligned nucleotide sequences. We define two a priori independent changepoint processes that describe variation in phylogenetic tree topologies and evolutionary pressures respectively. This approach results in more accurate recombination detection over the single MCP model, where spatial phylogenetic variation is modeled with one changepoint process. We demonstrate the increased accuracy of recombination detection by simulating recombination at different sites in the real dataset of mtDNA sequences from four primates. Sampling from the posterior distribution is accomplished via reversible jump MCMC. To simultaneously increase the precision of the recombination identification, one can use informative priors on recombination locations. We develop a new class of priors using a spatial point process that allows incorporating information about previously detected recombinants a flexible manner. We apply these priors to an HIV nucleotide sequence dataset to estimate the locations of recombination sites.
|
