Computer simulation of DNA sequential correlations

Abstract

ABSTRACT:To obtain new insight into the nature of the correlations in distributions of nucleotides in DNA sequences, an efficient algorithm is presented by means of which one can generate nucleotide sequences with prescribed correlational properties. It is demonstrated in the case of human T-cell receptor beta chain gene segment that by extracting a finite number of parameters from the natural sequence one can generate artificial sequences which reproduce the correct correlational properties. It is shown how the apparent long-range correlations, or-what is equivalent-the 1rf form of the spectral density result due to the arrangement of DNA segments with different functional roles which also differ in composition in terms of nucleotide classes. This approach enables one to get under control the impact of statistical fluctuations which represent a major obstacle in the analyses of finite natural sequences.