Discriminating coding, non-coding and regulatory regions using rescaled range and detrended fluctuation analysis

Abstract

In this paper we analyse the efficiency of two methods, rescaled range analysis and detrended fluctuation analysis, in distinguishing between coding DNA, regulatory DNA and non-coding non-regulatory DNA of Drosophila melanogaster. Both methods were used to estimate the degree of sequential dependence (or persistence) among nucleotides. We found that these three types of DNA can be discriminated by both methods, although rescaled range analysis performs slightly better than detrended fluctuation analysis. On average, non-coding, non-regulatory DNA has the highest degree of sequential persistence. Coding DNA could be characterised as being anti-persistent, which is in line with earlier findings of latent periodicity. Regulatory regions are shown to possess intermediate sequential dependency. Together with other available methods, rescaled range and detrended fluctuation analysis on the basis of a combined purine/pyrimidine and weak/strong classification of the nucleotides are useful tools for refined structural and functional segmentation of DNA. (C) 2007 Elsevier Ireland Ltd. All rights reserved.