Related Experiment Videos
Computational selection of distinct class- and subclass-specific gene expression signatures.
Pierre R Bushel1, Hisham K Hamadeh, Lee Bennett
1National Institute of Environmental Health Sciences, P.O. Box 12233, Research Triangle Park, NC 27709, USA. bushel@niehs.nih.gov
Journal of Biomedical Informatics
|April 3, 2003
Summary
This study uses statistical methods to identify gene expression patterns in rat liver samples. These patterns effectively classify samples treated with enzyme inducers versus peroxisome proliferators, distinguishing subclasses of drugs.
Area of Science:
- Pharmacogenomics
- Computational Biology
- Toxicology
Background:
- Gene expression profiling is crucial for understanding cellular responses to xenobiotics.
- Distinguishing between different classes of chemical agents based on their molecular effects is a key challenge in toxicology.
Purpose of the Study:
- To develop and apply statistical methods for identifying gene expression signatures that classify biological samples.
- To differentiate between samples treated with an enzyme inducer (phenobarbital) and various peroxisome proliferators.
Main Methods:
- Utilized modified Z-score tests and binomial distribution for identifying differentially expressed genes.
- Applied hierarchical clustering to group samples based on gene expression profiles.
- Employed analysis of variance (ANOVA) and linear discriminant analysis for gene selection and classification.
Main Results:
- Identified 238 statistically significant differentially expressed genes.
- Successfully partitioned samples into distinct groups based on treatment (enzyme inducer vs. peroxisome proliferator).
- Discerned subclasses within peroxisome proliferator-treated samples (fibrate subclass).
Conclusions:
- Statistical analysis of gene expression data can effectively categorize subclasses of samples exposed to pharmacologic agents.
- Proposed a classification regimen combining replicate data analysis, outlier diagnostics, and gene selection for microarray data.
- This approach aids in understanding drug mechanisms and subclassifications.