Title: Utilizing Machine Learning to Greatly Expand the Range and Accuracy of Bottom-Up Coarse-Grained Models Through Virtual Particles Show Chinese title

Title: 利用机器学习通过虚拟粒子极大地扩展自下而上粗粒化模型的范围和准确性

Abstract: Coarse-grained (CG) models parameterized using atomistic reference data, i.e., 'bottom up' CG models, have proven useful in the study of biomolecules and other soft matter. However, the construction of highly accurate, low resolution CG models of biomolecules remains challenging. We demonstrate in this work how virtual particles, CG sites with no atomistic correspondence, can be incorporated into CG models within the context of relative entropy minimization (REM) as latent variables. The methodology presented, variational derivative relative entropy minimization (VD-REM), enables optimization of virtual particle interactions through a gradient descent algorithm aided by machine learning. We apply this methodology to the challenging case of a solvent-free CG model of a 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) lipid bilayer and demonstrate that introduction of virtual particles captures solvent-mediated behavior and higher-order correlations which REM alone cannot capture in a more standard CG model based only on the mapping of collections of atoms to the CG sites. Show Chinese abstract

Abstract: 基于原子参考数据的粗粒度（CG）模型，即“自下而上”的CG模型，在生物分子和其他软物质的研究中已被证明是有用的。 然而，构建高度准确、低分辨率的生物分子CG模型仍然是一个挑战。 在本工作中，我们展示了如何在相对熵最小化（REM）的框架内，将没有原子对应关系的虚拟粒子作为潜在变量引入CG模型中。 所提出的的方法，变分导数相对熵最小化（VD-REM），通过机器学习辅助的梯度下降算法实现了对虚拟粒子相互作用的优化。 我们将这种方法应用于一个无溶剂的CG模型，即1,2-二油酰-sn-甘油-3-磷酸胆碱（DOPC）脂质双层，并证明引入虚拟粒子能够捕捉到溶剂介导的行为和高阶相关性，而仅基于原子集合到CG位点映射的标准CG模型无法通过REM单独捕捉这些特性。