A Computational Tool for Modeling of Protein Macromolecules Guided by Cryo-Electron Microscopy Data

摘要 冷冻电子显微镜（cryo-EM）是一种强大的技术，可以产生大分子的体积图像。以近原子（<5 μ m）分辨率产生的图像可用于确定这些分子的结构。由于实验的困难，只有一小部分的图像产生在近原子的分辨率，而主要数量的可用图像产生在亚/纳米分辨率。在亚纳米（5 - 10纳米）分辨率下，结构的骨架不能直接从这些图像中构建。然而，从头建模可以用来推导分子的原子结构。从体积图像中检测二级结构元素（螺旋和片层）对于从头工具至关重要。此外，观察这些元素之间的结构连接是非常有帮助的，以回答拓扑确定问题。蛋白质分子的拓扑结构确定问题可以定义为蛋白质分子序列上的二级结构元素与其冷冻电镜体积图像上的二级结构元素之间的对应关系。这个问题被证明是NP难的。在这个项目中，将开发一个完整的从头系统，能够有效地从真实的冷冻EM体积图像中导出大分子的结构。 拟议的从头系统被分为若干组件，每个组件将解决一个重要的子问题。从头建模将通过三个主要子系统来完成：（1）从其嘈杂的低温EM图像中提取分子的精细质量骨架;（2）解决拓扑确定问题;（3）构建目标分子的原子结构。提取的蛋白质分子骨架将被用来面对拓扑确定的问题，从而可以大大减少搜索空间的大小。此外，骨架将用于构建分子的结构。该系统将使用真实和合成图像的基准进行评估。 新的算法，将在建议的从头系统开发的各个领域都有意义。 例如，所提出的骨架化算法可以由使用骨架的生物医学科学使用。此外，本研究的动态匹配算法可以推广到任何研究领域的类似匹配问题。此外，所提出的系统将有助于理解一些蛋白质类型的基本功能，如膜蛋白和大蛋白质分子，这些蛋白质很难使用传统的实验技术进行研究。 该项目将带来合作者的努力和技能，来自田纳西州立大学（大津）计算机科学系的Kamal Al Nasr博士，Wei Chen博士和Matthew Hayes博士，以及来自弗吉尼亚联邦大学（VCU）的顾问Montserrat Samso博士，以及本科生和研究生。主要研究人员将负责开发拟议的算法，并进行项目的设计和分析部分。学生将接触到生物信息学研究，他们将获得一些重要的生物信息学问题的实践经验。学生在主要研究人员的指导下，将收集数据，实施所提出的算法，并分析结果。此外，该项目将帮助大津培养一支强大的少数民族学生队伍，他们可以在生物信息学的各个领域与工业界或学术界的同行竞争。建议的活动将加强生物学和计算机科学的研究和教育的融合，预计本科生将更有动力从事医学领域的职业。

Abstract Cryo-Electron Microscopy (cryo-EM) is a powerful technique produces volumetric images of large molecules. The images produced at near-atomic (<5Å) resolution can be used to determine the structure of those molecules. Due to experimental difficulties, only small portion of the images are produced at near-atomic resolution while the dominant number of available images is produced at sub/nanometer resolution. At subnanometer (5-10Å) resolution, the backbone of the structure cannot be constructed directly from those images. Nevertheless, de novo modeling can be used to derive the atomic structure of the molecules. The detection of secondary structure elements (helices and sheets) from the volumetric images is crucial for de novo tools. Moreover, the observation of the structural connections between these elements is extremely helpful in order to answer the topology determination problem. Topology determination problem can be defined as the correspondence between the secondary structure elements found on the sequence of the protein molecule and those found on its cryo-EM volumetric image. This problem is proven to be NP-hard. In this project, a complete de novo system that is capable of efficiently deriving the structure of large molecules from the authentic cryo-EM volumetric images will be developed. The proposed de novo system is divided into a number of components each of which will address an important sub-problem. De novo modeling will be accomplished by three main sub-systems (1) extracting a fine-quality skeleton of the molecule from its noisy cryo-EM image (2) addressing the topology determination problem and (3) building the atomic structure of the target molecule. The extracted skeleton of the protein molecule will be used to confront the problem of topology determination so that the search space size can be drastically reduced. In addition, the skeleton will be used to construct the structure of the molecule. The system will be evaluated using a benchmark of authentic and synthesized images. The novel algorithms that will be developed in the proposed de novo system are significant to various fields. For example, the proposed algorithm of skeletonization can be used by biomedical sciences that use the skeletons. Moreover, the dynamic matching algorithm of this research can be generalized to be used for similar matching problems in any field of study. Further, the proposed system will help understanding the fundamental functions of some protein types such as membrane proteins and large protein molecules that are hard to study using the traditional experimental techniques. This project will bring the effort and skills of the collaborators, Dr. Kamal Al Nasr, Dr. Wei Chen, and Dr. Matthew Hayes from the Department of Computer Science at Tennessee State University (TSU), and the consultant, Dr. Montserrat Samso from Virginia Commonwealth University (VCU), with the undergraduate and graduate students. The principal investigators will be responsible to develop the proposed algorithms and carry out the design and analysis part of the project. The students will be exposed to bioinformatics research and they will gain hands-on experience with some important bioinformatics problems. The students, with the guidance of the principle investigators, will collect data, implement the proposed algorithms, and analyze the results. Further, this project will help TSU prepare a strong workforce of minority students who can compete with their peers in industry or academia in various areas of bioinformatics. The proposed activities will enhance integration of research and education in biology and computer science and it is expected that undergraduate students will be more motivated to pursue career in medical fields.