Periodic power spectrum with applications in detection of latent periodicities in DNA sequences

TL;DR

This paper introduces a novel method for detecting latent periodicities in DNA sequences by directly computing the periodic power spectrum, which outperforms traditional Fourier-based approaches in capturing complex genomic periodicities.

Contribution

The paper presents a new technique to compute the full periodic power spectrum of DNA sequences directly from nucleotide distributions, enhancing detection of latent periodicities.

Findings

Method captures broader latent periodicities in genomes.

Outperforms conventional Fourier transform in accuracy.

Minimizes spectral leakage effects.

Abstract

Latent periodic elements in genomes play important roles in genomic functions. Many complex periodic elements in genomes are difficult to be detected by commonly used digital signal processing (DSP). We present a novel method to compute the periodic power spectrum of a DNA sequence based on the nucleotide distributions on periodic positions of the sequence. The method directly calculates full periodic spectrum of a DNA sequence rather than frequency spectrum by Fourier transform. The magnitude of the periodic power spectrum reflects the strength of the periodicity signals, thus, the algorithm can capture all the latent periodicities in DNA sequences. We apply this method on detection of latent periodicities in different genome elements, including exons and microsatellite DNA sequences. The results show that the method minimizes the impact of spectral leakage, captures a much broader latent periodicities in genomes, and outperforms the conventional Fourier transform.