Winger Moritz, Trzesniak Daniel, Baron Riccardo, van Gunsteren Wilfred F
Laboratory of Physical Chemistry, Swiss Federal Institute of Technology, ETH, 8093, Zürich, Switzerland.
Phys Chem Chem Phys. 2009 Mar 28;11(12):1934-41. doi: 10.1039/b818713d. Epub 2009 Feb 18.
The use of a coarse-grained (CG) model that is widely used in molecular dynamics simulations of biomolecular systems is investigated with respect to the dependence of a variety of quantities upon the size of the used integration time step and cutoff radius. The results suggest that when using a non-bonded interaction-cutoff radius of 1.4 nm a time step of maximally 10 fs should be used, in order not to produce energy sinks or wells. Using a too-large time step, e.g. 50 fs with a cutoff of 1.2 nm, as is done in the coarse-grained model of Marrink et al. (J. Phys. Chem. B, 2004, 108, 250 and 2007, 111, 7812), induces errors due to the linear approximation of the integrators that are commonly used to integrate the equations of motion. As a spin-off of the investigation of the mentioned CG models, we found that the parameters of the CG water model place it at physiological temperatures well into the solid phase of the phase diagram.
针对各种量对所使用的积分时间步长大小和截止半径的依赖性,研究了在生物分子系统分子动力学模拟中广泛使用的粗粒度(CG)模型。结果表明,当使用1.4 nm的非键相互作用截止半径时,为了不产生能量阱或势阱,应使用最大为10 fs的时间步长。使用过大的时间步长,例如在Marrink等人的粗粒度模型(《物理化学杂志B》,2004年,108卷,250页和2007年,111卷,7812页)中使用1.2 nm截止值时采用50 fs,由于通常用于对运动方程进行积分的积分器的线性近似会导致误差。作为对上述CG模型研究的附带成果,我们发现CG水模型的参数使其在生理温度下处于相图的固相区域。