Order and correlations in genomic DNA sequences. The spectral approach

The structural analysis of genomic DNA sequences is discussed in the framew ork of the spectral approach, which is sufficiently universal due to the re ciprocal correspondence and mutual complementarity of Fourier transform len gth scales. The spectral characteristics of random sequences of the same nu cleotide composition possess the property of self-averaging for relatively short sequences of length M greater than or equal to 100-300. The compariso n with the characteristics of random sequences determines the statistical s ignificance of the structural features observed. Apart from traditional app lications to the search of hidden periodicities, spectral methods are also efficient in studying mutual correlations in DNA sequences. By combining sp ectra for structure factors and correlation functions, not only integral co rrelations can be estimated but also their origin identified. Using the str uctural spectral entropy approach, the regularity of a sequence can be quan titatively assessed. A brief introduction to the problem is also presented and other major methods of DNA sequence analysis described.