粗粒化分子动力学模拟常常被用于解释实验观察的现象、预测自组装形貌以及聚集行为，但是常常面临精确性较差的问题。一个主要的问题是，不同粗粒化珠子之间的交叉项相互作用没有被合适地参数化。这项工作提出了一种新型的自上而下和自下而上相结合策略来对自身和交叉项相互作用进行参数化。选择的体系是水溶液中的磷脂。采用了分段Morse势来描述非键范德华相互作用。其中，自身相互作用的参数通过重现实验密度、蒸发焓、以及表面张力以一种自上而下的方式来优化。交叉项相互作用的参数通过重现全原子模拟得到的假性质（包括混合密度、分子间作用能、以及径向混合系数）以一种自下而上的方式来优化。这套粗粒化力场的可移植性通过重现磷脂膜在液相和凝胶相中的大量的结构性质和热力学性质得以保证。这套力场可以很好地预测了囊泡自组装和囊泡融合的过程。匹配假性质打开了发展高精度和高可移植性粗粒化力场的新途径。

Coarse-graining (CG) molecular dynamics (MD) simulations are widely used in interpreting experimental observations and predicting assembly morphology as well as collective behaviour but also face the problem of poor accuracy. A main issue is that cross-termed interactions between different CG beads are inadequately parameterized. This work proposes a novel top-down and bottom-up combined strategy to parameterize both self- and cross-termed interactions of zwitterionic phospholipids in water solution based on a piecewise Morse potential describing nonbonded van der Waals interactions. The self-interacting force parameters were optimized by matching experimental density, heat vapourization, and surface tension in a top-down manner, while the cross-termed interactions were optimized by fitting pseudo properties obtained from atomistic simulations in a bottom-up way, including mixing density, intermolecular energy, and radial mixing coefficient. The transferability of the CG force field (FF) was confirmed by reproducing a variety of structural and thermodynamic properties of lipid membranes in both liquid and gel phases. This FF can well depict vesicle self-assembly and vesicle fusion processes. Matching pseudo properties opens a new way to develop CG FF with increased accuracy and transferability.