|
Activity Number:
|
95
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 4, 2003 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #301361 |
|
Title:
|
Analysis of Cdna Microarray Data from Split-plot Experiments Using Mixed Models
|
|
Author(s):
|
Hao Li*+ and Arnold J. Stromberg and Thomas V. Getchell and Marilyn L. Getchell
|
|
Companies:
|
University of Kentucky and University of Kentucky and University of Kentucky and University of Kentucky
|
|
Address:
|
1608 University Court, Lexington, KY, 40503,
|
|
Keywords:
|
split-plot design ; mixed model ; microarray
|
|
Abstract:
|
Split-plot designs are utilized increasingly in clinical and biomedical research. However the statistical approach has not been fully developed for analyzing cDNA microarray data from split-plot experiments. We present a basic framework for analyzing this type of microarry data using a two-step mixed model. Such analysis allows us to quantify various sources of error and determine the optimal sample sizes based on the research paradigm. Since microarray experiments are frequently exploratory in nature, the genes with significant interactions or main effects may both be of interest. We construct an approximate overall F-test to select differentially expressed genes and then further categorize them based on whether the interaction terms are significant or not. We summarize and categorize the interactions with simple diagrams. We demonstrate our method on the analysis of microarray data from two regions of the brain from control subjects and patients with Alzheimer's disease and compare our approach to other commonly used methods of analysis. Supported by NIH AG-16345 (MLG).
|
