|
Activity Number:
|
357
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #307902 |
|
Title:
|
Hierarchical Modeling and Stochastic Variable Selection in the Analysis of Multiple SNPs in Genetic Association Studies
|
|
Author(s):
|
David Conti*+ and Juan P. Lewinger
|
|
Companies:
|
University of Southern California and University of Southern California
|
|
Address:
|
1501 San Pablo Street, ZNI 445, Los Angeles, CA, 90089,
|
|
Keywords:
|
hierarchical modeling ; genetic association ; haplotype ; stochastic variable selection ; SNP ; Bayesian analysis
|
|
Abstract:
|
With numerous SNPs available within a candidate gene region, one often evaluates both SNP- and haplotype-level associations. In addition to multiple comparison issues, conventional estimation may lead to unstable and biased estimates due to sparse data. Here, we jointly model the SNPs and introduce a modified interaction term to capture the underlying haplotype structure. This analysis estimates both the risk associated with each variant and the importance of phase between pairwise combinations of SNPs. To avoid unstable estimation due to sparse data, we propose a combination hierarchical model and stochastic variable selection procedure to highlight key SNPs and phase terms while incorporating uncertainty in model selection. We demonstrate the performance of this method under various genetic scenarios using simulations and discuss an application to real data.
|
