Bulletin of the American Physical Society
APS March Meeting 2017
Volume 62, Number 4
Monday–Friday, March 13–17, 2017; New Orleans, Louisiana
Session A25: Advances in Molecular Dynamics Simulations: From Atomistic to Coarse Grained ModelsIFocus

Hide Abstracts 
Sponsoring Units: DCP Chair: Gregory Voth, University of Chicago Room: 288 
Monday, March 13, 2017 8:00AM  8:36AM 
A25.00001: Entropy as a collective variable Invited Speaker: Michele Parrinello Sampling complex free energy surfaces that exhibit long lived metastable states separated by kinetic bottlenecks is one of the most pressing issues in the atomistic simulations of matter. Not surprisingly many solutions to this problem have been suggested. Many of them are based on the identification of appropriate collective variables that span the manifold of the slow varying modes of the system. While much effort has been put in devising and even constructing on the fly appropriate collective variables there is still a cogent need of introducing simple, generic, physically transparent, and yet effective collective variables. Motivated by the physical observation that in many case transitions between one metastable state and another result from a trade off between enthalpy and entropy we introduce appropriate collective variables that are able to represent in a simple way these two physical properties. We use these variables in the context of the recently introduced variationally enhanced sampling and apply it them with success to the simulation of crystallization from the liquid and to conformational transitions in protein. [Preview Abstract] 
Monday, March 13, 2017 8:36AM  8:48AM 
A25.00002: Machine Learning of Quantum Forces: building accurate force fields for molecular dynamics simulation via ``covariant'' kernels. Aldo Glielmo, Peter Sollich, Alessandro De Vita In recent years, Machine Learning algorithms have proven successful in the construction of datadriven force fields that bridge the gap between accurate (but slow) quantum chemical calculations and the fast (but unreliable) classical interatomic potentials. Such schemes learn either the local energy of a specific atom [Behler et al. PRL (2007), Bart\'{o}k et al. PRL, 2010] or its relative force [Li et al. PRL, 2015]. Within Learn On The Fly (LOTF) [Cs\'{a}nyi et al. PRL, 2004] simulations, the second approach is particularly suited since it guarantees reference accuracy on database entries. I will discuss a novel scheme [Glielmo et al. PRB, submitted] to accurately predict atomic forces as vector quantities, rather than sets of scalar components, by Gaussian Process (GP) Regression. This is based on matrixvalued kernel functions, to which we impose that the predicted force rotates with the target configuration and is independent of any rotations applied to the configuration database entries. We show that such ``covariant'' GP kernels can be obtained by integration over the elements of the rotation group SO(n). The accuracy of our kernels in predicting quantum forces in real materials is investigated by tests on pure and defective Ni and Fe crystalline systems. [Preview Abstract] 
Monday, March 13, 2017 8:48AM  9:00AM 
A25.00003: AMOEBA 2.0: A physicsfirst approach to biomolecular simulations Joshua Rackers, Jay Ponder The goal of the AMOEBA force field project is to use classical physics to understand and predict the nature of interactions between biological molecules.~ While making significant advances over the past decade, the ultimate goal of predicting binding energies with ``chemical accuracy'' remains elusive.~ The primary source of this inaccuracy comes from the physics of how molecules interact at short range.~ For example, despite AMOEBA's advanced treatment of electrostatics, the force field dramatically overpredicts the electrostatic energy of DNA stacking interactions.~ AMOEBA 2.0 works to correct these errors by including simple, first principles physicsbased terms to account for the quantum mechanical nature of these shortrange molecular interactions.~ We have added a charge penetration term that considerably improves the description of electrostatic interactions at short range.~ We are reformulating the polarization term of AMOEBA in terms of basic physics assertions.~ And we are reevaluating the van der Waals term to match \textit{ab initio} energy decompositions.~ These additions and changes promise to make AMOEBA more predictive.~ By including more physical detail of the important shortrange interactions of biological molecules, we hope to move closer to the ultimate goal of true predictive power. [Preview Abstract] 
Monday, March 13, 2017 9:00AM  9:36AM 
A25.00004: Multiscale simulations of patchy particle systems combining Molecular Dynamics, Path Sampling and Green's Function Reaction Dynamics Invited Speaker: Peter Bolhuis Important reactiondiffusion processes, such as biochemical networks in living cells, or selfassembling soft matter, span many orders in length and time scales. In these systems, the reactants' spatial dynamics at mesoscopic length and time scales of microns and seconds is coupled to the reactions between the molecules at microscopic length and time scales of nanometers and milliseconds. This wide range of length and time scales makes these systems notoriously difficult to simulate. While meanfield rate equations cannot describe such processes, the mesoscopic Green's Function Reaction Dynamics (GFRD) method enables efficient simulation at the particle level provided the microscopic dynamics can be integrated out. Yet, many processes exhibit nontrivial microscopic dynamics that can qualitatively change the macroscopic behavior, calling for an atomistic, microscopic description. The recently developed multiscale Molecular Dynamics Green's Function Reaction Dynamics (MDGFRD) approach combines GFRD for simulating the system at the mesocopic scale where particles are far apart, with microscopic Molecular (or Brownian) Dynamics, for simulating the system at the microscopic scale where reactants are in close proximity. The association and dissociation of particles are treated with rare event path sampling techniques. I will illustrate the efficiency of this method for patchy particle systems. Replacing the microscopic regime with a Markov State Model avoids the microscopic regime completely. The MSM is then precomputed using advanced pathsampling techniques such as multistate transition interface sampling. I illustrate this approach on patchy particle systems that show multiple modes of binding. MDGFRD is generic, and can be used to efficiently simulate reactiondiffusion systems at the particle level, including the orientational dynamics, opening up the possibility for largescale simulations of e.g. protein signaling networks. [Preview Abstract] 
Monday, March 13, 2017 9:36AM  9:48AM 
A25.00005: Prediction of purification of biopharmeceuticals with molecular dynamics Vincent Ustach, Roland Faller Purification of biopharmeceuticals remains the most expensive part of proteinbased drug production. In ion exchange chromatography (IEX), prediction of the elution ionic strength of host cell and target proteins has the potential to reduce the parameter space for scaleup of protein production. The complex shape and charge distribution of proteins and pores complicates predictions of the interactions in these systems. Allatom molecular dynamics methods are beyond the scope of computational limits for mass transport regimes. We present a coarsegrained model for proteins for prediction of elution pH and ionic strength. By extending the raspberry model for colloid particles to surface shapes and charge distributions of proteins, we can reproduce the behavior of proteins in IEX. The average charge states of titratatable amino acid residues at relevant pH values are determined by extrapolation from allatom molecular dynamics at pH 7. The pH specific allatom electrostatic field is then mapped onto the coarsegrained surface beads of the raspberry particle. The hydrodynamics are reproduced with the latticeBoltzmann scheme. This combination of methods allows very long simulation times. The model is being validated for known elution procedures by comparing the data with experiments. [Preview Abstract] 
Monday, March 13, 2017 9:48AM  10:00AM 
A25.00006: Molecular Dynamics based lattice gas automata Alexander Wagner, Reza Parsa We present a lattice gas (LG) model derived from an underlying Molecular Dynamics (MD) simulation. In principle any MD simulation will result in a corresponding MDLG model. The question is then for which cases can we derive LG collision operators that only depend on the current state of the LG system. We show that for systems that approximate an ideal gas we can recover the standard lattice Boltzmann algorithm to good approximation. We conclude with an outlook for extending this approach to derive coarse grained lattice gas models for fluctuating dynamics and nonideal systems. [Preview Abstract] 
Monday, March 13, 2017 10:00AM  10:36AM 
A25.00007: Coarsegrained molecular dynamics simulations for giant proteinDNA complexes. Invited Speaker: Shoji Takada Biomolecules are highly hierarchic and intrinsically flexible. Thus, computational modeling calls for multiscale methodologies. We have been developing a coarsegrained biomolecular model where onaverage 1020 atoms are grouped into one coarsegrained (CG) particle (1). Interactions among CG particles are tuned based on atomistic interactions and the fluctuation matching algorithm. CG molecular dynamics methods enable us to simulate much longer time scale motions of much larger molecular systems than fully atomistic models. After broad sampling of structures with CG models, we can easily reconstruct atomistic models, from which one can continue conventional molecular dynamics simulations if desired. Here, we describe our CG modeling methodology for proteinDNA complexes, together with various biological applications, such as the DNA duplication initiation complex, model chromatins, and transcription factor dynamics on chromatinlike environment. (1) Takada, S.; Kanada, R.; Tan, C.; Terakawa, T.; Li, W.; Kenzaki,~H. Modeling Structural Dynamics of Biomolecular Complexes by CoarseGrained Molecular Simulations~Accounts of Chemical Research~48: 30263035, 2015. (2) Shimizu, M.; Noguchi, Y.;, Sakiyama, Y.; Kawakami, H.; Katayama, T.; Takada, S. Nearatomic structural model for bacterial DNA replication initiation complex and its functional insights. Proc. Nat. Acad. Sci. USA in press. [Preview Abstract] 
Monday, March 13, 2017 10:36AM  10:48AM 
A25.00008: Probing the Phase Behavior of CoarseGrained Polymer Models with Nested Sampling Kenneth Salerno, Noam Bernstein The phase behavior of polymers is not as well studied as that of atomic systems due to the highly correlated motion of polymers and resulting sampling difficulties. Nested sampling (NS) is a statistical technique that allows calculation of the partition function of physical systems by eliminating a fixed fraction of configuration space at each iteration of the algorithm. Previous studies have shown that by using NS one can directly calculate thermodynamic quantities such as heat capacity of atomic systems from the partition function. We report results from recent work extending NS to polymeric systems using Hamiltonian and Galilean Monte Carlo sampling methods. Beadspring models of flexible and semiflexible singlechain systems that exhibit a coil to globule transition are studied. Results for thermodynamic quantities, such as the heat capacity, and chain structural quantities, like $R_g$, are presented. Results from physically based coarsegrained and atomistic polyethylene models are also discussed. Altogether, these results show how NS can be applied to calculate polymer phase behavior in a computationally efficient way. [Preview Abstract] 
Monday, March 13, 2017 10:48AM  11:00AM 
A25.00009: Dynamics in entangled polyethylene melts using coarsegrained models Brandon L. Peters, Gary S. Grest, K. Michael Salerno, Anupriya Agrawal, Dvora Perahia Polymer dynamics creates distinctive viscoelastic behavior as a result of a coupled interplay of motion on multiple length scales. Capturing the broad time and length scales of polymeric motion however, remains a challenge. Using polyethylene (PE) as a model system, we probe the effects of the degree of coarse graining on polymer dynamics. Coarsegrained (CG) potentials are derived using iterative Boltzmann inversion (iBi) with 26 methyl groups per CG bead from all fully atomistic melt simulations for short chains. While the iBi methods produces nonbonded potentials which give excellent agreement for the atomistic and CG pair correlation functions, the pressure P = 100500MPa for the CG model. Correcting for potential so P ~ 0 leads to nonbonded models with slightly smaller effective diameter and much deeper minimum. However, both the pressure and nonpressure corrected CG models give similar results for mean squared displacement (MSD) and the stress auto correlation function G(t) for PE melts above the melting point. The time rescaling factor between CG and atomistic models is found to be nearly the same for both CG models. Transferability of potential for different temperatures was tested by comparing the MSD and G(t) for potentials generated at different temperatures. [Preview Abstract] 
Follow Us 
Engage
Become an APS Member 
My APS
Renew Membership 
Information for 
About APSThe American Physical Society (APS) is a nonprofit membership organization working to advance the knowledge of physics. 
© 2020 American Physical Society
 All rights reserved  Terms of Use
 Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 207403844
(301) 2093200
Editorial Office
1 Research Road, Ridge, NY 119612701
(631) 5914000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 200452001
(202) 6628700