DSP based entropy estimation for identification and classification of Homo sapiens cancer genes

Because of advancement of microarray technology in recent years, public domains like NCBI, NIH, NHGRI etc. present extensive range of information rich raw genomic data. Effortless accessibility of these data attracts the researchers from diverse disciplines to process them for the benefit of the society. In the field of signal processing, a new area of research has been introduced namely genomic signal processing (GSP). GSP basically processes genes, proteins and DNA sequences using various signal processing methodologies to extract the information hidden in it. As some genetic abnormalities turn into cancer diseases, proper understanding and analysis of genes and proteins may lead to a new horizon in cancer genomic study. In genomic signal processing, exact identification and classification of diseased gene is a great challenge to the researchers. Hence in the present paper, the crucial job of gene identification and classification is attempted. As a solution to this problem, statistical methods like entropy estimation and mutual information calculation is adopted along with DSP technique. Rayleigh distribution of estimated entropy of gene is treated as identifier of healthy and cancerous Homo sapiens. Once the cancer genes are identified, mutual information estimator based on their minimum entropy is used as classifier to detect different types of cancer genes. The present algorithms are successfully tested on several healthy and cancerous prostate, breast and colon genes collected from NCBI genbank.