Hierarchical Coarse Graining Via A Generalizedyvon Born Green Framework

Download or read book Hierarchical Coarse-graining Via a Generalizedyvon-born-green Framework written by Joseph Rudzinski and published by . This book was released on 2014 with total page pages. Available in PDF, EPUB and Kindle. Book excerpt: Atomically-detailed molecular dynamics simulations have emerged as one of the most powerful theoretictools for studying complex, condensed-phase systems. Despite their ability to provide incredible molecularinsight, these simulations are insufficient for investigating complex biological processes, e.g., protein foldingor molecular aggregation, on relevant length and time scales. The increasing scope and sophistication ofatomically-detailed models has motivated the development of "hierarchical" approaches, which parameterizea low resolution, coarse-grained (CG) model based on simulations of an atomically-detailed model. Theutility of hierarchical CG models depends on their ability to accurately incorporate the correct physics ofthe underlying model. One approach for ensuring this "consistency" between the models is to parame-terize the CG model to reproduce the structural ensemble generated by the high resolution model. Themany-body potential of mean force is the proper CG energy function for reproducing all structural distri-butions of the atomically-detailed model, at the CG level of resolution. However, this CG potential is aconfiguration-dependent free energy function that is generally too complicated to represent or simulate. Themultiscale coarse-graining (MS-CG) method employs a generalized Yvon-Born-Green (g-YBG) relation todirectly determine a variationally optimal approximation to the many-body potential of mean force. TheMS-CG/g-YBG method provides a convenient and transparent framework for investigating the equilibriumstructure of the system, at the CG level of resolution. In this work, we investigate the fundamental limitationsand approximations of the MS-CG/g-YBG method. Throughout the work, we propose several theoretic con-structs to directly relate the the MS-CG/g-YBG method to other popular structure-based CG approaches.We investigate the physical interpretation of the MS-CG/g-YBG correlation matrix, the quantity responsiblefor disentangling the various contributions to the average force on a CG site. We then employ an iterativeextension of the MS-CG/g-YBG method that improves the accuracy of a particular set of low order correla-tion functions relative to the original MS-CG/g-YBG model. We demonstrate that this method provides apowerful framework for identifying the precise source of error in an MS-CG/g-YBG model. We then proposea method for identifying an optimal CG representation, prior to the development of the CG model. Weemploy these techniques together to demonstrate that in the cases where the MS-CG/g-YBG method failsto determine an accurate model, a fundamental problem likely exists with the chosen CG representation orinteraction set. Additionally, we explicitly demonstrate that while the iterative model successfully improvesthe accuracy of the low order structure, it does so by distorting the higher order structural correlationsrelative to the underlying model. Finally, we apply these methods to investigate the utility of the MS-CG/g-YBG method for developing models for systems with complex intramolecular structure. Overall, our resultsdemonstrate the power of the g-YBG framework for developing accurate CG models and for investigatingthe driving forces of equilibrium structures for complex condensed-phase systems. This work also explicitlymotivates future development of bottom-up CG methods and highlights some outstanding problems in thefield.