Signal detection in genome sequences using complexity based features

Abstract

In this work, we tackle the problem of evaluating complexity methods and measures for finding interesting signals in the whole genome of three prokaryotic organisms. In addition to previous complexity measures, new measures are introduced for representing Open Reading Frames (ORF). We apply different classification algorithms to determine which complexity measure results in better predictive performance in discriminating genes from pseudo-genes in ORFs. Also, we investigate whether positions and lengths of windows in ORFs have significant impact on distinguishing between genes and pseudo-genes. Different classification algorithms are applied for classifying ORFs into genes and pseudo-genes.