Random fields approach to the study of DNA chains

We apply the random field theory to the study of DNA chains which we assume to be trajectories of a stochastic process. We construct statistical potential between nucleotides corresponding to the probabilities of those trajectories that can be obtained from the DNA data base containing millions of sequences. It turns out that this potential has an interpretation in terms of quantities naturally arrived at during the study of evolution of species i. e. probabilities of mutations of codons. Making use of recently performed statistical investigations of DNA we show that this potential has different qualitative properties in coding and noncoding parts of genes. We apply our model to data for various organisms and obtain a good agreement with the results just presented in the literature. We also argue that the coding/noncoding boundaries can corresponds to jumps of the potential.