|
Activity Number:
|
233
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 3, 2009 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #304262 |
|
Title:
|
Impact of Covariates on Feature Selection in Microarray Analysis
|
|
Author(s):
|
Elizabeth McClellan*+ and Monnie McGee and Richard Scheuermann
|
|
Companies:
|
Southern Methodist University and Southern Methodist University and Pathology U.T. Southwestern Medical Center
|
|
Address:
|
Department of Statistical Science, Dallas, TX, 75275,
|
|
Keywords:
|
microarray analysis ; feature selection ; covariates ; classification
|
|
Abstract:
|
A necessary application of the DNA microarray process is the development of analytical methods for feature selection, a tool that determines which genes are differentially expressed in various types of tissues. However, measured covariates that are not of interest may confound the relationship between expression values from varying tissue types such that a list of significant genes may contain truly uninformative genes. If such covariates affect gene expression, their impact should be eliminated to minimize frequencies of false gene discoveries. The need to incorporate covariates into gene expression analysis is examined here by regressing gene expression on covariates and a treatment to determine significantly differentially expressed genes based on the p-value of the treatment in the model. Results indicate that modeling gene expression with covariates may not always be necessary.
|
