A clustering algorithm for gene expression data using wavelet packet decomposition

Mining large amounts of data and deriving meaning from the mined data in bioinformatics is a computationally intensive and relevant issue. In this paper, an efficient algorithm to cluster genes into similar 'functional' groups is presented. This is a technique for extracting and characterizing rhythmic expression profiles from genome-wide DNA micro-array hybridization data. These patterns are clues to discovering rhythmic genes implicated in cell-cycle, circadian, or other biological processes. These functionalities are discussed. A signal-processing approach to this problem is presented. The information theoretic criterion for identifying those genes exhibiting maximum variation in behavior is explored. The genes are clustered and then relationships are derived for the proposition of a temporal cell-cycle model governing regulatory behavior. The human fibroblast and yeast data set are presently considered for analysis.