Random fields approach to the study of DNA chains.

ABSTRACT

We apply the random field theory tothe study of DNA chains which we assume tobe trajectories of a stochastic process. Weconstruct statistical potential betweennucleotides corresponding to theprobabilities of those trajectories thatcan be obtained from the DNA data basecontaining millions of sequences. It turnsout that this potential has aninterpretation in terms of quantitiesnaturally arrived at during the study ofevolution of species i.e. probabilities ofmutations of codons. Making use of recentlyperformed statistical investigations of DNAwe show that this potential has differentqualitative properties in coding andnoncoding parts of genes. We apply ourmodel to data for various organisms andobtain a good agreement with the resultsjust presented in the literature. We alsoargue that the coding/noncoding boundariescan corresponds to jumps of the potential.