Variations in base pair composition and associated long-range correlations in DNA sequences--computer simulation results.

Abstract

Recently, the possible occurrence of long-range correlations between nucleotides in DNA sequences of living organisms has excited considerable interest. Of particular importance is the claim that only intron-containing sequences exhibit these correlations. Different investigations, however, have disproved the claimed difference between intron-containing and intron-less DNA sequences. Moreover, very recent investigations pointed out that the long-range correlations appear only if relatively large variations in nucleotide composition along the DNA sequence are present. Furthermore, some examples demonstrated that these variations may have clear biological reasons. In this paper we investigate in detail, with the aid of computer simulations, the connection between compositional heterogeneity of a DNA sequence and the appearance of long-range correlations. As an explicit example, the DNA sequence of the lambda-phage is compared with different artificial sequences of similar compositional heterogeneity. The results demonstrate that the variations of the nucleotide composition along the sequence can fully account for all properties of the claimed long-range correlations. New results of extensive computer simulations are presented which clearly demonstrate how the apparent 'fractal' (or 'long-range-correlated') character of a sequence gradually evaporates, as the frequency of the compositional variations of a simulated sequence continuously increases.