Wavelet transforms for the analysis of microarray experiments

Abstract

Array comparative genomic hybridization (cgh) is a microarray technology for measuring the relative copy number of thousands of genomic regions. Visual examination of cgh profiles shows that genomic changes occur on a variety of length scales. Such changes may be characteristic of phenotypic variables such as tumor type and gene mutational status. To aid in identifying such features and exploring their relationship with phenotypic outcomes, we are applying wavelet transforms to the analysis of such profiles. This allows us to decompose a cgh signal into components on different length scales, even when the genome is severely aberrated, providing a convenient basis for exploring their behavior. Wavelet transforms may also be useful in the realm of gene expression. The expression signal given by genes in clustered order can be wavelet transformed, which compresses the signal from many genes into a few components, possibly aiding in the development of new tumor classifiers.