|
Activity Number:
|
59
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Sunday, July 29, 2007 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #310259 |
|
Title:
|
Bayesian Multivariate Spatial Models for Association Mapping in Structured Samples
|
|
Author(s):
|
Meijuan Li*+ and Brad Carlin and Cavan Reilly
|
|
Companies:
|
The University of Minnesota and The University of Minnesota and The University of Minnesota
|
|
Address:
|
A448 Mayo Bldg., MMC 303, 420 Delaware St. SE, Minneapolis, MN, 55455-0378,
|
|
Keywords:
|
Population Structure ; Association Mapping ; Bayesian Conditional Autoregressive (CAR) Modeling ; Linkage-Disequilibrium ; Relative Kinship ; Deviance Information Criterion (DIC)
|
|
Abstract:
|
Population-based Linkage-Disequilibrium (LD) mapping permits much finer-scale mapping and higher power than does family linkage analysis. However, unlike family based association study, the population-based association mapping is not a controlled experiment, false positives can arise from population structure and genetic familiar relatedness between the samples. There is tremendous interest in simultaneously testing the association between a candidate gene and multiple phenotypes of interest. We here present a new method for population-based association mapping by multivariate Bayesian conditional autoregressive modeling (MCAR). The method we developed accounts for population structure and complex relationships between the samples. We illustrate our modeling approach using the previously published 4 type of flowering data from 95 Arabidopsis thaliana accessions.
|
