Classification of Microarray Data to Predict Toxic Exposure

by Seleem, Tarek A.

Seleem, Tarek A. M.S., Department of Computer Science, Wright State University, 2007. Classification of Microarray Data to Predict Toxic Exposure. This thesis presents a software system for the analysis of microarray data. Microarrays are a relatively new technology that can be used to examine the state of the genome of an organism at some instant in time. The challenge is the amount of natural variation in biological systems limits our ability to identify specific genes that may be sensitive to changes in an organism’s physiology or its environment. The analysis software consists of three modules. The first module filters microarray data to reduce the complexity of the problem. The second module selects subsets of genes for evaluation using a genetic algorithm. The final module uses a neural network to evaluate the selected genes to predict an organism’s level of exposure to toxic substance. Results are present for a data set consisting of subjects exposed to eight levels of ?-naphthylisothiocyanate (ANIT), a model hepatotoxin causing liver damage in the form of intrahepatic cholestasis.