Prediction of compound signature using high density gene expression profiling

Area of Science:

• Toxicogenomics
• Molecular Toxicology
• Bioinformatics

Background:

• DNA microarrays enable simultaneous measurement of thousands of gene expressions.
• Gene expression profiles linked to known compound exposures can form a predictive database.
• Previous work demonstrated cDNA microarrays generate chemical-specific gene expression profiles.

Purpose of the Study:

• To test if knowledge about blinded samples can be gained using a gene expression profile database.
• To validate the hypothesis that gene expression profiling aids in predicting toxicological effects.
• To assess the utility of gene expression databases for hazard identification.

Main Methods:

• A training set of rat liver gene expression profiles from exposures to known compounds was established.
• Highly discriminant genes were identified using linear discriminant analysis (LDA) and genetic algorithm/K-nearest neighbors (GA/KNN).
• These genes were used to analyze blinded liver RNA samples from exposures to phenytoin, diethylhexylpthalate, or hexobarbital.

Main Results:

• Successful prediction of whether blinded samples originated from rats exposed to enzyme inducers or peroxisome proliferators.
• Demonstrated the ability to distinguish between different classes of chemical exposures based on gene expression.
• Validated the use of gene expression profiles for classifying unknown toxicological exposures.

Conclusions:

• The study validates the hypothesis that gene expression databases can provide knowledge about unknown chemical exposures.
• Further development of gene expression databases will significantly enhance toxicological hazard identification processes.
• This approach holds promise for efficient screening of therapeutic drugs and understanding chemical effects.