Developing a new protein homology-modeling method with high precision

点击翻译按钮获取中文摘要

This study aimed to develop a protein-structure modeling method that can reproduce the native 3D structure of a target protein starting from a model structure obtained by the usual homology-modeling method. Employing the latest computational theories, the resulting full-atom protein model structure would be as precise as that determined by the X-ray crystallography (i.e., 2A or better in the atomic resolution). In order to realize this sort of method, there are two main problems to be solved. One is a so-called force-field problem, in which the X-ray structure does not correspond to the minimum energy point when the conformation energy is calculated with the existing force fields. The other is a sampling problem in the structural space of a protein, where energy minimization should be carried out overcoming high energy barriers as going one minimum point to another.For the first problem above, a statistical potential system was developed from the database analysis for backbone dihedral … More angles of all 20 amino acids, and this potential system was incorporated into the AMBER force fields in place of the corresponding energy terms. The modified AMBER system provided significantly better results in terms of the minimum energy point relative to the X-ray structure. For the second problem, we extended the usual multicanonical theory by employing the Wang-Landau theory. Replacing the Monte Carlo procedure originally used in the latter theory with the molecular dynamics, we developed a Wang-Landau Molecular Dynamic (WL MD) technique, which enabled automatic estimation of weighting factors during the molecular simulation. In this way, the difficulty in the conventional multicanonical method was overcome, and fast and efficient search in the structural space was realized. The new method was applied to folding simulation of a small-sized protein (Trp-Cage), in comparison with that by the standard replica exchange MD (RE-MD) method. The results showed clearly better efficiency with the WL-MD method than that with RE-MD. Less