Stochastic tunneling minimization by molecular dynamics: an application to heteropolymer models

We report the results of an application of the stochastic tunneling (STUN) minimization method to two models of linear heteropolymers. One is a protein model of Honeycutt and Thirumalai and the other is a model of polyampholyte with specific charge distribution. By using molecular dynamics simulations, we examined low-energy conformational space of the two models by means of standard simulated annealing (SA) and stochastic tunneling techniques. We find that the STUN method consistently outperforms the regular SA algorithm in the simulations of the two model systems studied. In the case of the protein, the stochastic tunneling method is able to locate about 5 times more low-energy conformations at short simulation times than the SA method while for the polyampholyte the achieved acceleration is about 2 times only. Possible reasons for the observed difference in performance of the stochastic tunneling algorithm are discussed