A gene selection method for classifying cancer samples using 1D discrete wavelet transform.

Selecting a set of discriminant genes for biological samples is an important task for designing highly efficient classifiers using DNA microarray data. The wavelet transform is a very common tool in signal-processing applications, but its potential in the analysis of microarray gene expression data is yet to be explored fully. In this paper, we present a wavelet-based feature selection method that assigns scores to genes for differentiating samples between two classes. The gene expression signal is decomposed using several levels of the wavelet transform. The genes with the highest scores are selected to form a feature set for sample classification. In this study, the feature sets were coupled with k-nearest neighbour (kNN) classifiers. The classification accuracies were assessed using several real data sets. Their performances were compared with several commonly used feature selection methods. The results demonstrate that 1D wavelet analysis is a valuable tool for studying gene expression patterns.