生物大分子多尺度理论和计算方法

分子模拟是研究生物大分子的重要手段. 过去二十年来, 人们将分子模拟与实验研究相结合, 揭示出生物大分子结构和动力学方面的诸多重要性质. 传统分子模拟主要采用全原子分子模型或各种粗粒化的分子模型. 在实际应用中, 传统分子模拟方法通常存在精度或效率瓶颈, 一定程度上限制了其应用范围. 近年来, 多尺度分子模型越来越受到人们的关注. 多尺度分子模型基于统计力学原理, 将全原子模型和粗粒化模型相耦合, 有望克服传统分子模拟方法中的精度/效率瓶颈, 进而拓展分子模拟在生物大分子研究中的应用范围. 根据模型之间的耦合方式, 近年来发展起来的多尺度分子模拟方法可归纳为如下四种类型: 混合分辨多尺度模型、并行耦合多尺度模型、单向耦合多尺度模型、以及自学习多尺度模型. 本文将对上述四类多尺度模型做简要介绍, 并讨论其主要优缺点、应用范围以及进一步发展方向.     

基于TorchMD的粗粒化分子动力模拟研究

粗粒化模型通过简化原子性质以及原子间的相互作用实现生物大分子长时间尺度的分子动力学模拟. 深度学习通过模拟人类的认知过程实现海量数据的准确分类和回归过程. 本论文将这两种技术进行融合,利用基于深度学习的粗粒化分子动力学模拟技术研究分子在不同状态之间的变化过程,并提出基于TorchMD的分子动力学模拟的分析框架. 在本工作中,MFDP聚类算法被用于在三维的CV变量空间中进行聚类,并确定分子的若干主要状态,在完成聚类的同时,给出各类中的代表分子构象,并给出类之间的分子构象. 这为后续利用String算法分析分子在不同状态间的转换路径打下基础. 通过String算法,迭代搜索得到分子在不同状态之间的变化路径以及对应的势能变化曲线. 通过与已有文献的结果进行对比,验证了基于TorchMD的粗粒化分子动力学模拟的理论框架可以在相对较短的时间尺度里研究分子的变化过程.     

Protein folding kinetics and thermodynamics from atomistic simulations

Determining protein folding kinetics and thermodynamics from all-atom molecular dynamics (MD) simulations without using experimental data represents a formidable scientific challenge because simulations can easily get trapped in local minima on rough free energy landscapes. This necessitates the computation of multiple simulation trajectories, which can be independent from each other or coupled in some manner, as, for example, in the replica exchange MD method. Here we present results obtained with a new analysis tool that allows the deduction of faithful kinetics data from a heterogeneous ensemble of simulation trajectories. The method is demonstrated on the decapeptide Chignolin for which we predict folding and unfolding time constants of 1.0 +/- 0.3 and 2.6 +/- 0.4 micros, respectively. We also derive the energetics of folding, and calculate a realistic melting curve for Chignolin.     

In silico folding of a three helix protein and characterization of its free-energy landscape in an all-atom force field

We report the reproducible first-principles folding of the 40 amino-acid, three-helix headpiece of the HIV accessory protein in a recently developed all-atom free-energy force field. Six of 20 simulations using an adapted basin-hopping method converged to better than 3 A backbone rms deviation to the experimental structure. Using over 60 000 low-energy conformations of this protein, we constructed a decoy tree that completely characterizes its folding funnel.     

Atomistic simulations of a thermotropic biaxial liquid crystal

We have performed molecular dynamics simulations of a 2,5-bis-(p-hydroxyphenyl)-1,3,4-oxadiazole mesogen (ODBP-Ph-C(7)) at a fully atomistic level for a range of temperatures within the region that has experimentally been assigned to a biaxial nematic phase. Analysis of the data shows that the simulated nematic phase is biaxial but that the degree of biaxiality is small. The simulations show also the formation of ferroelectric domains in the nematic where the molecular short axis is aligned with the oxadiazole dipoles parallel to each other. Removal of electrostatic interactions leads to destabilization of ferroelectric domains and destabilization of the biaxiality. An additional simulation shows the slow growth of a mesophase directly from the isotropic fluid over a period of approximately 50 ns. This is the first time this has been achieved within the framework of an all-atom model.     

A network of conformational transitions in an unfolding process of HP-35 revealed by high-temperature MD simulation and a Markov state model

An understanding of protein folding/unfolding processes has important implications for all biological processes, including protein degradation, protein translocation, aging, and diseases. All-atom molecular dynamics(MD) simulations are uniquely suitable for it because of their atomic level resolution and accuracy. However, limited by computational capabilities, nowadays even for small and fast-folding proteins, all-atom MD simulations of protein folding still presents a great challenge. An alternative way is to study unfolding process using MD simulations at high temperature. High temperature provides more energy to overcome energetic barriers to unfolding, and information obtained from studying unfolding can shed light on the mechanism of folding. In the present study, a 1000-ns MD simulation at high temperature(500 K)was performed to investigate the unfolding process of a small protein, chicken villin headpiece(HP-35). To infer the folding mechanism, a Markov state model was also built from our simulation, which maps out six macrostates during the folding/unfolding process as well as critical transitions between them, revealing the folding mechanism unambiguously.     

A study of Barstar folding events using boundary value simulations

From extensive biophysical studies of protein folding, two competing mechanisms emerged: hydrophobic collapse and the framework model. Our protein of choice is Barstar—a barnase inhibitor. The approximation algorithm we used to study Barstar folding trajectories is called SDEL—stochastic difference equation in length. Using the native structure as the final boundary value and a collection of unfolded structures as the varying initial boundary value, SDEL calculates an ensemble of least action pathways between these boundaries. The results are atomically detailed folding pathways, with as many intermediate structures as you request in the input. We generated 12 pathways, starting from a structurally wide selection of unfolded conformations. Using the protein's radius of gyration as our primary reaction coordinate, we tracked H-bonds, dihedral angles, native and non-native contacts, and energy along the folding pathways. This paper will follow our findings, with special emphasis on pinpointing hydrophobic collapse as a more appropriate mechanism for Barstar. Comparison with pathway predictions for Barstar using experimental techniques will also be discussed.     

Mechanical unfolding and refolding pathways of ubiquitin

Mechanical unfolding and refolding of ubiquitin are studied by Monte Carlo simulations of a Gō model with binary variables. The exponential dependence of the time constants on the force is verified, and folding and unfolding lengths are computed, with good agreement with experimental results. Furthermore, the model exhibits intermediate kinetic states, as observed in experiments. Unfolding and refolding pathways and intermediate states, obtained by tracing single secondary structure elements, are consistent with simulations of previous all-atom models and with the experimentally observed step sizes.     

Simple theory of protein folding kinetics

We present a simple model of protein folding dynamics that captures key qualitative elements recently seen in all-atom simulations. The goals of this theory are to serve as a simple formalism for gaining deeper insight into the physical properties seen in detailed simulations as well as to serve as a model to easily compare why these simulations suggest a different kinetic mechanism than previous simple models. Specifically, we find that non-native contacts play a key role in determining the mechanism, which can shift dramatically as the energetic strength of non-native interactions is changed. For proteinlike non-native interactions, our model finds that the native state is a kinetic hub, connecting the strength of relevant interactions directly to the nature of folding kinetics.     

Quantitative protein dynamics from dominant folding pathways

We develop a theoretical approach to the protein-folding problem based on out-of-equilibrium stochastic dynamics. Within this framework, the computational difficulties related to the existence of large time scale gaps are removed, and simulating the entire reaction in atomistic details using existing computers becomes feasible. We discuss how to determine the most probable folding pathway, identify configurations representative of the transition state, and compute the most probable transition time. We perform an illustrative application of these ideas, studying the conformational evolution of alanine dipeptide, within an all-atom model based on the empiric GROMOS96 force field.     

Molecular dynamics simulations of membrane deformation induced by amphiphilic helices of Epsin,Sar1p,and Arf1

The N-terminal amphiphilic helices of proteins Epsin, Sar1p, and Arf1 play a critical role in initiating membrane deformation. The interactions of these amphiphilic helices with the lipid membranes are investigated in this study by combining the all-atom and coarse-grained simulations. In the all-atom simulations, the amphiphilic helices of Epsin and Sar1p are found to have a shallower insertion depth into the membrane than the amphiphilic helix of Arf1, but remarkably,the amphiphilic helices of Epsin and Sar1p induce higher asymmetry in the lipid packing between the two monolayers of the membrane. The insertion depth of amphiphilic helix into the membrane is determined not only by the overall hydrophobicity but also by the specific distributions of polar and non-polar residues along the helix. To directly compare their ability to deform the membrane, the coarse-grained simulations are performed to investigate the membrane deformation under the insertion of multiple helices.     

Beyond finite-size scaling in solidification simulations

Although computer simulation has played a central role in the study of nucleation and growth since the earliest molecular dynamics simulations almost 50 years ago, confusion surrounding the effect of finite size on such simulations has limited their applicability. Modeling solidification in molten tantalum on the Blue Gene/L computer, we report here on the first atomistic simulation of solidification that verifies independence from finite-size effects during the entire nucleation and growth process, up to the onset of coarsening. We show that finite-size scaling theory explains the observed maximal grain sizes for systems up to about 8 000 000 atoms. For larger simulations, a crossover from finite-size scaling to more physical size-independent behavior is observed.     

Generating ensemble averages for small proteins from extended conformations by Monte Carlo simulations

Since it is not feasible to determine the structure of every protein by experiment, algorithms delivering the folded conformation of a protein solely from its amino acid sequence are desirable. Here the diffusion-process controlled-Monte Carlo approach has been applied to generating ensemble averages for three small proteins with 31, 36, and 46 residues. Starting from extended conformations and using an energy model that was developed on other protein models, the simulations find nativelike structures deviating by 3 A rms from the experimental structures for the main chain atoms. The balance between long-range and short-range interactions is discussed briefly in the context of stability and folding.     

Challenges in protein folding simulations: Timescale, representation, and analysis

Experimental studies of protein folding processes are frequently hampered by the fact that only low resolution structural data can be obtained with sufficient temporal resolution. Molecular dynamics simulations offer a complementary approach, providing extremely high resolution spatial and temporal data on folding processes. The effectiveness of such simulations is currently hampered by continuing questions regarding the ability of molecular dynamics force fields to reproduce the true potential energy surfaces of proteins, and ongoing difficulties with obtaining sufficient sampling to meaningfully comment on folding mechanisms. We review recent progress in the simulation of three common model systems for protein folding, and discuss how recent advances in technology and theory are allowing protein folding simulations to address their current shortcomings.     

Embedding atom-jellium model for metal surface

To describe metal surfaces efficiently and accurately, an embedding atom-jellium model is proposed. Within density functional theory, we consider a multiscale scheme that combines jellium and atomistic approaches. We use the former to model layers deep inside a metal surface to reduce the computational cost and the later to maintain the accuracy required for chemical bonding. Work functions of Al(111) and Cu(111) surfaces are studied using this model with comparisons to all-atom and pure jellium models. The much closer results of the embedding atom-jellium model to the all-atom results than to the pure jellium results show a good prospect for our approach in large-scale density functional calculations.     

SiC晶界薄膜的变电荷分子动力学模拟

基于迭代变电荷方法,用分子动力学模拟了SiC中的晶界薄膜．从原子尺度上模拟了不同的晶界薄膜的结构．观察到了晶粒与晶界薄膜间的电荷转移并且晶界薄膜的厚度与电荷转移有关．该结果提供了晶界存在空间电荷的直接证据,并证明静电作用与晶界薄膜的平衡厚度密切相关． 关键词： 分子动力学 变电荷 晶界薄膜     

Proper orthogonal decomposition of atomistic flow simulations

An adaptive proper orthogonal decomposition based on time windows (WPOD) for analysis of velocity fields from atomistic simulations is presented. The method effectively separates the field into ensemble average and fluctuation components, and can be applied to both stationary and non-stationary flows in simple and complex geometries. The criteria to distinguish POD modes representing ensemble average and fluctuation components are based on adaptive examination of eigenvalue decomposition, specifically on the rate of decay of POD eigenvalues and analysis of POD eigenvectors. The WPOD method is efficient and its superior accuracy leads to smooth field gradients that can be used effectively in multiscale formulations. The method has been applied in molecular dynamics and dissipative particle dynamics simulations. We demonstrate the method for several cases including steady and unsteady flow and red blood cell flow, but the same approach can be used in atomistic simulations of materials.     

Modeling hydrogen exchange of proteins by a multiscale method

We proposed a practical way for mapping the results of coarse-grained molecular simulations to the observables in hydrogen change experiments. By combining an atomic-interaction based coarse-grained model with an all-atom structure reconstruction algorithm, we reproduced the experimental hydrogen exchange data with reasonable accuracy using molecular dynamics simulations. We also showed that the coarse-grained model can be further improved by imposing experimental restraints from hydrogen exchange data via an iterative optimization strategy. These results suggest that it is feasible to develop an integrative molecular simulation scheme by incorporating the hydrogen exchange data into the coarse-grained molecular dynamics simulations and therefore help to overcome the accuracy bottleneck of coarse-grained models.     

双壁碳纳米管电浸润现象的分子动力学模拟

在单壁碳纳米管电浸润现象原子模拟的基础上,对双壁碳纳米管的电浸润现象进行了计算机模拟.运用经典分子动力学方法结合一个宏观的电毛细管模型模拟了双壁碳纳米管在水银中的电浸润过程,对不同内管尺寸情况下的浸润现象作了研究和比较.计算结果表明双壁碳管和单壁碳管的电浸润过程存在很大的不同,双壁碳管的内管在电浸润过程中起到重要的作用：当改变双壁碳管中内管的尺寸时,浸润现象会产生很大的改变. 关键词： 双壁碳纳米管 电浸润 分子动力学