ON THE ROBUSTNESS OF MAXIMUM-ENTROPY RELATIONSHIPS FOR COMPLEXITY DISTRIBUTIONS OF NUCLEOTIDE-SEQUENCES

Given a functionally equivalent set of natural nucleotide sequences, t he distribution of local compositional complexity among all subsequenc es of this set appears to be as random as possible consistent with the mean complexity of such subsequences. The robustness of this relation ship and its possible causes have been explored by means of (1) dynami c simulations based on models of biased substitution mutations, (2) eq uilibrium models incorporating known mononucleotide probabilities, and (3) extension of the analyses, previously carried out on short oligon ucleotides, to much larger subsequences. The maximum entropy effect ev idently follows from almost any mechanism for substitution mutation dy namics that incorporates a systematic bias toward low-complexity. The effect is only partially explained by unequal mononucleotide probabili ties. The complexity distributions for larger subsequences of length r ange 40-120 nucleotides show novel regularity of structure ('featherin g') that is not yet explained.