Empirical model of residue contact probabilities for polypeptides

An empirical theory for the probability of formation of intrachain contacts in denatured polypeptides is presented. Contact probabilities for arbitrary sequences are modeled by composition-weighted averages of the homopolypeptides contact probabilities. Homopolypeptide probabilities are obtained for each one of the 20 amino acids by Monte Carlo simulations. A multipower law model for the homopolypeptides fits very precisely the contact probabilities after optimization. Its asymptotic behavior for large contact loop lengths consist of three different exponents depending on where the contacts are located along the chain in general agreement with other theoretical models. The model for the heteropolypeptide contact probabilities is also able to reproduce the simulated probabilities of a set of denatured proteins but with less accuracy. Contact probabilities for heteropolypeptides are significantly similar, in particular for loop lengths above 20 residues, and the model does not have enough resolution to clearly distinguish between different proteins. Sequence order information seems necessary to increase the precision. Overall, contacts with loop lengths above 33 residues behave as those from random flight chains. One exception are polyproline chains for which this length seems to be much higher.