Cancer classification using entropy analysis in fractional Fourier domain of gene expression profile

The vast advancement in the field of DNA microarrays has enabled researchers to simultaneously analyse the expression levels of thousands of genes on a single microarray chip. Several data-mining methods have been applied in studying the gene expression profiles to distinguish between various sub-types of cancer and types of other diseases. However, accurate diagnosis of cancer sub-types remains a challenge. The gene-by-gene-based approaches are likely to produce chance correlations owing to the high-dimensional nature of the microarray experiments, and the fact that biological phenomena are constantly cyclical and rhythmic. Gene expression is highly regulated and correlations exist between the expressions of different genes; therefore, the cooperativity of genes must be taken into consideration to capture the inherent characteristics of the genome. The present method utilized fractional Fourier transform (FRFT) and entropy-based techniques to extract representative features of the gene expression profile (GEP) and conduct tumour classification using the support vector machine (SVM). The algorithm was tested using four different data-sets. The experimental results reveal that this algorithm has the ability to classify cancers into various types and sub-types with high accuracy.