Development of rare-event sampling techniques for predicting structures and free energies of crystal polymorphs and oligopeptides

开发罕见事件采样技术来预测晶体多晶型物和寡肽的结构和自由能

• 批准号: 1565980
• 负责人: Mark Tuckerman
• 金额: $ 58万
• 依托单位: New York University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2016
• 资助国家: 美国
• 起止时间: 2016-08-15 至 2020-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1565980&HistoricalAwards=false
• 关键词: Development; rare; event; sampling; techniques

Mark Tuckerman of New York University is supported by an award from the Chemical Theory, Models and Computational Methods program in the Chemistry Division to develop methods and software for the prediction of molecular crystal structure. This award is cofunded by the CISE/ACI Software Reuse Venture Fund. In the science of materials, ordered arrays of molecules forming structures known as molecular crystals play an essential role in the pharmaceutical, electronics, and defense industries. Often, the crucial question is which crystals should be made for a particular application. It is worth noting that one of the most widely used pharmaceutical molecular crystals, aspirin, was discovered essentially by accident. Typically, in crystal engineering, it is necessary to screen large databases of potential candidate compounds. Unfortunately, making and characterizing molecular crystals in the laboratory is generally time consuming and costly, rendering a trial-and-error approach through such a database impractical. How many more important molecular crystal systems might be discovered if a systematic, targeted approach could be applied? Theory and computation, which can, in principle, rapidly predict molecular crystal structures and their properties, are uniquely poised to play a key role in creating such a targeted approach. What is needed, however, are robust algorithms for making these predictions. The Tuckerman group develops computational techniques and software for predicting the crystal structures a given compound can form and ranking them according to a thermodynamic property known as free energy, which has been recognized in the scientific community as the proper figure of merit for such a ranking but has remained an elusive property to determine. The Tuckerman group also adapts these algorithms for studying the conformational preferences of short chains of amino acids known as oligopeptides in order to explore the role these important biological molecules play in immunogenicity and the design of new classes of pharmaceuticals. Tuckerman and his coworkers are engaged in many software activities including developing a computer package for crystal structure prediction, improving the efficiency of their molecular dynamics software, PINY-MD and continuing to contribute software to many community software codes. All of the software developed in this project is made available to the broader research community. The basic properties of molecular materials in the solid state are often strongly influenced by the details of their crystal structures and the existence of polymorphs. Experimental determination of these structures is costly and time-consuming, which places increased importance on the role of theory and computation. Similarly, the biochemical function of small oligopeptides, from immunogenicity to inhibition, is affected by their equilibrium conformations in different environments. Computational prediction of structure in complex systems such as these is challenging due to the so-called rare-event sampling problem on a rough potential energy landscape, which arises when attempting to study the equilibrium thermodynamics and kinetics of many complex systems. Roughness on an energy surface refers to the existence of high barriers to conformational and structural changes. The Tuckerman group has proposed to develop robust free-energy based enhanced sampling algorithms and software for overcoming the rare-event problem that arises in the crystal structure prediction and conformational sampling of oligopeptides, thereby allowing favored structures to be identified and thermodynamically ranked in an efficient manner. In the proposed methods, the free energy landscape is expressed in terms of select set of collective variables (CVs) designed to distinguish the different structural motifs in these systems. The CVs are first be subject to new surface navigation techniques in order to identify the minima and saddles points, collectively referred to as "landmarks" on the landscape, and then targeted for enhanced sampling in order to produce the free energy ranking of the landmarks. The new techniques are applied to predict the crystal structures and polymorphs of both rigid and flexible small organic molecules, to study the conformational free energy landscape of an immunogenic peptide binding to the major histocompatibility complex, and to understand the influence of mechanical force on the unfolding mechanism of â-hairpin peptide. Software creation will be accelerated via hackathons organized by the Tuckerman group. Education of students in rare-event methods is aided through workshops organized at New York University's global campus sites. Finally, the Tuckerman group reaches out to underrepresented groups via national organizations having a presence in New York City in order to help devise and participate in STEM-related educational activities.

纽约大学的Mark Tuckerman获得了化学部化学理论、模型和计算方法项目的支持，以开发预测分子晶体结构的方法和软件。该奖项由CEISE/ACI软件重用风险基金共同资助。在材料科学中，形成分子晶体结构的有序分子阵列在制药、电子和国防工业中发挥着重要作用。通常，关键的问题是应该为特定的应用制造哪些晶体。值得注意的是，阿司匹林是最广泛使用的药物分子晶体之一，基本上是偶然发现的。通常，在晶体工程中，有必要筛选潜在候选化合物的大型数据库。不幸的是，在实验室中制造和表征分子晶体通常既耗时又昂贵，这使得通过这样一个数据库进行反复试验的方法不切实际。如果可以应用系统的、有针对性的方法，还可能发现多少更重要的分子晶体系统？理论和计算在原则上可以快速预测分子晶体结构及其性质，在创造这种有针对性的方法方面具有独特的优势。然而，我们需要的是做出这些预测的强大算法。塔克曼团队开发了计算技术和软件，用于预测给定化合物可以形成的晶体结构，并根据一种名为自由能的热力学性质对它们进行排名，自由能在科学界被认为是进行此类排名的适当优值，但仍然是一个难以确定的性质。塔克曼小组还采用这些算法来研究被称为寡肽的氨基酸短链的构象偏好，以探索这些重要的生物分子在免疫原性和设计新类别药物中所起的作用。Tuckerman和他的同事从事许多软件活动，包括开发用于晶体结构预测的计算机程序包，提高他们的分子动力学软件Piny-MD的效率，并继续为许多社区软件代码贡献软件。在这个项目中开发的所有软件都可供更广泛的研究社区使用。固态分子材料的基本性质往往受到其晶体结构的细节和多晶型的存在的强烈影响。这些结构的实验确定是昂贵和耗时的，这使得理论和计算的作用变得更加重要。同样，小分子寡肽的生化功能，从免疫原性到抑制作用，都受到不同环境中平衡构象的影响。在这样的复杂系统中，结构的计算预测是具有挑战性的，因为在粗略势能图景上出现了所谓的罕见事件采样问题，这是在试图研究许多复杂系统的平衡热力学和动力学时出现的。能量表面的粗糙度是指存在着阻碍构象和结构变化的高势垒。Tuckerman小组建议开发基于自由能的增强采样算法和软件，以克服在低聚肽的晶体结构预测和构象采样中出现的罕见事件问题，从而允许以有效的方式识别受欢迎的结构并进行热力学排序。在所提出的方法中，自由能图景是通过选择集合变量(CV)来表示的，该集合变量旨在区分这些系统中的不同结构基元。CV首先受到新的表面导航技术的影响，以便识别地貌上的最小点和鞍点，这些点统称为“地标”，然后针对目标进行增强采样，以便产生地标的自由能排名。这些新技术被应用于预测刚性和柔性有机小分子的晶体结构和晶型，研究免疫原肽与主要组织相容性复合体结合的构象自由能景观，以及了解机械力对发夹肽展开机制的影响。通过Tuckerman团队组织的黑客马拉松，将加速软件的创建。通过在纽约大学全球校园网站组织的研讨会，对学生进行罕见事件方法的教育。最后，Tuckerman小组通过在纽约市有存在的国家组织接触代表性不足的群体，以帮助设计和参与与STEM有关的教育活动。