|
Activity Number:
|
421
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #310089 |
|
Title:
|
Bayesian Nonparametric Clustering of Amino Acid Usage Profiles
|
|
Author(s):
|
Daniel Merl*+
|
|
Companies:
|
Cornell University
|
|
Address:
|
580 Malott Hall, Ithaca, NY, 14853,
|
|
Keywords:
|
Dirichlet process ; positive selection ; Bayesian nonparametrics
|
|
Abstract:
|
The study of heterogeneity in amino acid usage is central to the field of molecular evolution. Increased heterogeneity at individual amino acid sites can be the result of relaxation of constraint, or the result of recurrent adaptive evolution (positive selection). I describe a Dirichlet process mixture model for achieving Bayesian nonparametric clustering of amino acid sites according to their usage profiles. Identification of sites with high posterior probabilities of belonging to clusters representing increased heterogeneity levels is shown by simulation study to be useful for detecting possible targets of positive selection. Since the clustering induced by the Dirichlet process is based on an unknown number of mixture components, the method is appropriate for the analysis of protein coding data for which there is little prior information about the variation in selective pressure.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2007 program |
