State Key Laboratory of Precision Spectroscopy, Department of Physics, Institute of Theoretical and Computational Science, East China Normal University , Shanghai 200062, China.
J Phys Chem B. 2013 Dec 5;117(48):14885-93. doi: 10.1021/jp4080866. Epub 2013 Nov 25.
An effective polarizable bond (EPB) model has been developed for computer simulation of proteins. In this partial polarizable approach, all polar groups of amino acids are treated as polarizable, and the relevant polarizable parameters were determined by fitting to quantum calculated electrostatic properties of these polar groups. Extensive numerical tests on a diverse set of proteins (including 1IEP, 1MWE, 1NLJ, 4COX, 1PGB, 1K4C, 1MHN, 1UBQ, 1IGD) showed that this EPB model is robust in MD simulation and can correctly describe the structure and dynamics of proteins (both soluble and membrane proteins). Comparison of the computed hydrogen bond properties and dynamics of proteins with experimental data and with results obtained from the nonpolarizable force field clearly demonstrated that EPB can produce results in much better agreement with experiment. The averaged deviation of the simulated backbone N-H order parameter of the B3 immunoglobobulin-binding domain of streptococcal protein G from experimental observation is 0.0811 and 0.0332 for Amber99SB and EPB, respectively. This new model inherited the effective character of the classic force field and the fluctuating feature of previous polarizable models. Different from other polarizable models, the polarization cost energy is implicitly included in the present method. As a result, the present method avoids the problem of over polarization and is numerically stable and efficient for dynamics simulation. Finally, compared to the traditional fixed AMBER charge model, the present method only adds about 5% additional computational time and is therefore highly efficient for practical applications.
已为蛋白质的计算机模拟开发出一种有效的极化键（EPB）模型。在这种部分极化方法中，将氨基酸的所有极性基团视为极化基团，并通过拟合这些极性基团的量子计算静电性质来确定相关的极化参数。对多种蛋白质（包括 1IEP、1MWE、1NLJ、4COX、1PGB、1K4C、1MHN、1UBQ、1IGD）进行的广泛数值测试表明，该 EPB 模型在 MD 模拟中具有鲁棒性，并且可以正确描述蛋白质（可溶性和膜蛋白）的结构和动力学。将计算得到的蛋白质氢键性质和动力学与实验数据以及非极化力场的结果进行比较，清楚地表明 EPB 可以产生与实验更吻合的结果。与 Amber99SB 相比，模拟链球菌蛋白 G 的 B3 免疫球蛋白结合结构域的骨架 N-H 序参量的模拟值与实验观察值的平均偏差分别为 0.0811 和 0.0332。该新模型继承了经典力场的有效特征和以前的极化模型的波动特征。与其他极化模型不同，本方法将极化代价能量隐含地包含在内。因此，本方法避免了过度极化的问题，在动力学模拟中具有数值稳定性和效率。最后，与传统的固定 AMBER 电荷模型相比，本方法仅增加了约 5%的额外计算时间，因此非常适合实际应用。
