Detection of statistical periodicities in DNA by Conflict and Entropy Minimization Methods

Some commonly used methods for statistical analysis of DNA sequences start by mapping each nucleotide to a real or complex number and, subsequently, applying classical methods of spectral analysis to the numerical sequences obtained. Such mappings may introduce inexistent spectral peaks or attenuate others whose presence is known. Consequently, recent approaches try to reduce the dependence of DNA spectral estimation on the chosen mapping through concepts of optimization and information theory. In this paper we examine the advantages of using complex mappings combined with convex optimization algorithms (Conflict and Entropy Minimization Methods) on the selection of appropriate base values for detecting statistical periodicities in DNA. These methods are applied to real DNA sequences extracted from public databases (Genbank and Ensembl).