Spectral analysis for detecting protein coding regions based on a new numerical representation of DNA [articol]

Abstract

The major signal in coding regions of genomic sequences has a three-base periodicity. By proposing a new numerical representation for the DNA chain, our aim is to use spectral analysis for recognizing the coding regions of a gene. Since the peak at f=1/3 in the Fourier spectrum is a good discriminator of the coding potential of an intronless DNA strand, we utilized this feature within a sliding window in order to detect probable exons in a DNA sequence. Our technique is independent of training sets or existing database information, and thus can find general application.