Extension of A Bayesian Recombination Detection Model
By FANG FANG
Graduate Student, Iowa State University
Genetic recombination explains a considerable amount of genetic diversity in natural populations. Accurate detection of recombination in DNA/RNA virus sequences is important because recombination can affect virus diversity, pathogenesis, and ability to evade immune pressures and antiviral agents. Recombination can be inferred when the phylogenetic relationships for different regions of a sequence disagree. Various methods are available. However, most identification methods fall into a sequential testing trap, first using the data to determine recombinant structure (parental and crossover point-COP locations) and then using the same data to assess significance conditional on the optimal recombinant structure. To overcome this shortfall, a Bayesian multiple change-point model and corresponding software (OhBrother) have been developed that can be used to simultaneously infer recombination and recombinant structure for HIV-1 [Suchard et al., 2002].
The aim of present work is to extend the methodology to use more comprehensive summary measures of the parental sequences, and demonstrate the usefulness of this method extension on both simulation data and real data.