Using SVD and SVM methods for selection, classification, clustering and modeling of DNA microarray data

DNA microarray technology is the latest and the most advanced too] for parallel measuring of the activity and interactions of thousands of genes. This modern technology promises new insight into mechanisms of living systems, for example only two high-density oligonucleotide microarrays are sufficient to inspect the whole human genome. However, it provides unprecedented amount of data that require application of advanced computational methods. The appropriate choice of data analysis technique depends both on data and on goals of an experiment. In this paper we focus on two promising methods: singular value decomposition and support vector machines. We discuss the possibility of application of these methods for different purposes; particularly for clustering, classification, feature selection and modeling of dynamics of gene expression. We use for testing presented approaches existing data sets, which are widely available via Internet, and one new tumor/normal thyroid microarray data set. (C) 2004 Elsevier Ltd. All rights reserved.