Toward experimental quality protein structures: a synergistic approach

描述（由申请人提供）：了解蛋白质折叠和错误折叠的机制，准确预测蛋白质结构是科学界面临的两个重要挑战。详细了解导致天然和非天然状态形成的分子事件是充分阐明蛋白质折叠机制的基础。随着代表性蛋白质序列家族的高通量结构表征的快速发展，一个基本的需求是高度精确的计算方法，可以可靠地产生接近实验质量的结构模型，以利用结构基因组学的投资。这些方法的可用性将使蛋白质结构的精确建模成为可能，这将对包括生物技术、制药工业、药物发现和生命科学在内的一系列领域产生重大影响。Duan和Zhou建议将两组的优势与互补的专业知识结合起来，开发蛋白质结构建模和优化的计算方法，最终目标是产生高度准确和可靠的蛋白质结构预测方法，具有与实验技术相当的准确性。目的1：在第一个具体目标中，Duan和Zhou提出开发新的蛋白质结构精细的构象采样方法。将利用并进一步发展最近开发的“生长适应”方法，以便从大量的视角蛋白质结构集合中准确识别近天然结构。这些方法的进一步发展将促进结构的细化，这将有助于改进结构，使其与从实验技术中获得的结构相当精确。目的2:Duan和Zhou提出开发有效的自由能（评分）函数，用于精确的全原子蛋白质结构建模。这种新的评分函数是基于整合基于知识的统计势和基于全原子物理的力场的协同概念。此外，与统计势的比较将允许对力场参数和溶剂化模型进行关键评估。目的3:Duan和Zhou提出以FSD1、protein G和protein L及其各自的不同拓扑突变体为模型系统，研究蛋白质天然结构拓扑在蛋白质折叠中的作用；研究三级构造形成对二级构造的依赖关系。与实验的比较，包括对我们模型预测能力的直接测试，将是我们研究的一个组成部分，并将有助于对该方法进行仔细审查。

DESCRIPTION (provided by applicant): Understanding the mechanisms of protein folding and misfolding and accurately predicting protein structures are two of the important challenges facing the scientific community. Detailed knowledge of the molecular events leading to the formation of both native and non-native states are the basis for a full elucidation of the protein folding mechanisms. With the rapid progress facilitated by the high-throughput structural characterization of representative protein sequence families, an essential need is the highly accurate computational methods that can reliably generate near-experimental quality structural models for capitalizing on the investment in the structural genomics. Availability of such methods would enable accurate modeling of protein structures which would have significant impact on a range of fields including biotechnology, pharmaceutical industry, drug discovery, and life sciences in general. Duan and Zhou propose to combine the strengths of the groups with complementary expertise to develop computational methods for protein structure modeling and refinement with the ultimate goal to produce highly accurate and reliable methods for protein structure prediction that have comparable accuracy to experimental techniques. Aim 1: Duan and Zhou propose to develop novel conformational sampling method for protein structure refinement in the first specific aim. A recently developed "Grow-to-Fit" method will be utilized and further developed to enable accurate identification of the near-native structures from a large ensemble of perspective protein structures. Further development of the methods will facilitate structural refinement which will help to improve the structures to be comparably accurate as those obtained from experimental techniques. Aim 2: Duan and Zhou propose to develop effective free energy (scoring) functions for accurate all-atom modeling of protein structures. This novel scoring function is based on the synergistic concept of integrating both knowledge-based statistical potential and the all-atom physics-based force field. Furthermore, comparison to the statistical potential will allow critical assessment of the force field parameters and solvation models. Aim 3: Duan and Zhou propose to examine the roles of protein native structure topology in protein folding using FSD1, Protein G and Protein L and their respective topologically distinct mutants as the model systems; to study the dependence of tertiary structure formation on secondary structures. Comparison with experiments, including direct tests on the predictive ability of our model will be an integral part of our study and will be instrumental for a close scrutiny on the approach. PUBLIC HEALTH RELEVANCE: To understand the basic rules of life, how cell works, it is necessary to know the protein structures that are critically important to understand how they work. This proposal is motivated by the need to develop computational method to reliably predict protein structures based on the primary sequence. Because protein structures are also enormously useful in drug discovery, a potential impact of the proposed work in human health is in the area of development of novel therapeutics.