|
Activity Number:
|
146
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 4, 2008 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #302208 |
|
Title:
|
A Novel Multistage Approach To Classify Samples Across Multiple Gene Expression Experiments
|
|
Author(s):
|
Heather A. Adams*+ and Sandra L. Rodriguez-Zas and Bruce Southey
|
|
Companies:
|
University of Illinois at Urbana-Champaign and University of Illinois at Urbana-Champaign and University of Illinois at Urbana-Champaign
|
|
Address:
|
1207 West Gregory Drive, Urbana, IL, 61801,
|
|
Keywords:
|
Classification ; Discrimination ; Mixed model ; Microarray experiment ; Meta-analysis
|
|
Abstract:
|
Presented is a two-stage method to identify classifier biomarker genes that provide accurate classification of samples across experiments. In stage one, non-linear mixed effects models were used to identify classifier transcripts across studies using three approaches integrating information across experiments by transcript: a) overlap of transcripts with expression profiles significantly associated with classes, b) joint analysis of estimates of association between gene expression and classes per experiment and c) joint analysis of association between gene expression and classes across experiments. Stage two evaluated multi-transcript classification approaches using the identified transcripts. Selection criteria were evaluated using correct classification rate. The analysis is powerful to identify consistent reliable classifier biomarker genes across experiments.
|
