Abstract
A reduced model, which can fold both helix and sheet structures, is proposed to study the problem of protein folding. The goal of this model is to find an unbiased effective potential that has included the effects of water and at the same time can predict the three dimensional structure of a protein with a given sequence in reasonable time. For this purpose, rather than focusing on the real folding dynamics or full structural details at the atomic scale, we adopt the Monte Carlo method and the coarse-grained representation of the protein in which both side-chains and the backbones are replaced by suitable geometrical objects in consistent with the known structure. On top of the coarse-grained representation, our effective potential can be developed. Two new interactions, the dipole-dipole interactions and the local hydrophobic interactions, are introduced and are shown to be as crucial as the hydrogen bonds for forming the secondary structures. In particular, for the first time, we demonstrate that the resulting reduced model can successfully fold proteins with both helix and sheet structures without using any biased potential. Further analyses show that this model can also fold other proteins in reasonable accuracy and thus provides a promising starting point for the problem of protein folding.