A Method to Classify Microarray Data

DNA microarrays are a technology that is used for monitoring the expression of thousand genes simultaneously. Microarray data create new opportunities for biomedical sciences but also pose serious challenges for statistical analysis, due to their complexity. Inference of biological meaning from gene expression profiling involves the classification of genes whose expression is correlated to a given functional condition. To test the value of Correspondence Analysis for microarray data classification, we compared gene expression profiles of cells in the presence or absence of a dominant negative Myc mutant. As compared to other microarray analysis methods, this procedure allows the projection of a complex data set into an easily visualized, low dimensional space, in which association between variables (genes) and observations (experiments) can be more easily evaluated. Correspondence Analysis is a powerful technique for gene-expression data analysis, particularly when combined with other classification techniques.