Identification of coding and non-coding sequences in a complete genome using local Hölder exponent formalism and Multi-affinity analysis

Accurate prediction of genes in genomes has always been a challenging task for bioinformaticians and computational biologists. Therefore, the discovery of relations in coding and non-coding sequences has led to new perspectives in the understanding of the DNA sequences. This has motivated us to find new methods to distinguish coding and non-coding sequences. We first introduce a number sequence representation of DNA sequences. Multi-affinity analysis and local Hölder exponent are then performed on the representation of the obtained number sequence. Three suited exponents are selected to form a parameter space. The two exponents γ(-2), γ(6) are from Multi-affinity analysis, the exponent h is from local Hölder exponent. Thus, each coding or non-coding sequence may be represented by a point in the three-dimensional parameter space. We can see the points corresponding to coding and non-coding sequences in the complete genome of many prokaryotes be divided to different regions roughly. If the point (γ(-2), γ(6), h) for a DNA sequence is situated in the region corresponding to coding sequences, the sequence is discriminated as a coding sequence; otherwise, the sequence is classified as a non-coding one. Therefore these exponents can be used to distinguish coding and non-coding sequences. The Fisher´s discriminant algorithm is used to give the discriminant accuracies. The average discriminant accuracies p_c, p_nc, q_c and q_nc of all 51 prokaryotes obtained by the present method reach 69.08%, 83.34%, 72.08% and 83.54%, respectively.