|
Activity Number:
|
23
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, July 29, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #309837 |
|
Title:
|
A Ridge Penalized Principal-Components Approach Based on Heritability for High-Dimensional Data
|
|
Author(s):
|
Yuanjia Wang*+ and Yixin Fang and Man Jin
|
|
Companies:
|
Columbia University and Columbia University and Columbia University
|
|
Address:
|
1531 9th St, Fort Lee, NJ, 07024,
|
|
Keywords:
|
Principal-component analysis ; Ridge penalty ; Heritability ; Family data ; Cross validation ; High-dimensional data
|
|
Abstract:
|
Microarray technique allows measurement of thousands of gene expression levels simultaneously and provides opportunity for mapping shared genetic contribution to multiple expression levels. Clustering analysis and principal component analysis are proposed to reduce the dimensionality of phenotypes. Genetic linkage or association analysis is then applied to the combined phenotypes. However, the usual clustering and principal component analysis are only appropriate for independent data. In most of the genetic studies where family data are available, applying these standard analyses to founders leads to loss of information. Here are proposed a clustering and a principal component approach based on heritability for high-dimensional data that take into account of the family structure information. The methods are illustrated through an application to human lymphoblastoid gene expression data.
|
