Compressive Structural BioInformatics: High Efficiency 3D Structure Compression

 描述(申请人提供)：蛋白质数据库(PDB)档案自2008年以来增加了一倍，2014年超过100,000个条目。与此同时，结构的大小和复杂性正在急剧增加，例如最近确定的艾滋病毒衣壳结构包含约250万个原子。综合结构生物学的新兴技术开始通过结合冷冻电子显微镜、小角X射线散射、X射线和核磁共振以越来越高的分辨率来确定兆道尔顿范围内的分子机器的结构。大型建筑群的交互式可视化超过了典型科学家台式机、笔记本电脑或移动设备的可用网络带宽和内存。对档案进行大规模的结构分析和查询已成为大数据的挑战。为了让所有科学家、教育工作者和学生都能访问这些结构，需要用新的方式来表示这些数据。在高清晰度电视、卫星通信、视频或音频流等领域，高效压缩一直是向手机、平板电脑、笔记本电脑和台式机提供交互式媒体的关键。在处理全基因组序列数据方面也出现了类似的趋势。一个完整的学科“压缩基因组学”已经被开发来处理数据压缩和处理这些数据的算法的开发。该建议引入了“压缩结构生物信息学”的概念，这是一套压缩算法、应用程序和工作流，以前所未有的速度(100-1000倍的加速比)分析和可视化大型结构和大型结构集，并以最小的客户端开销进行分析。该项目的目标是：1.开发紧凑和可扩展的3-D生物分子结构表示；2.通过减少网络带宽和支持数据流，实现大型复合体的交互可视化；3.实现对受I/O限制的工作流的PDB归档的大规模分析；以及4.开发开放源代码软件库。通过与广泛使用的可视化应用程序和分布式数据并行工作流系统的开发人员合作，将实施新技术，并对其进行基准测试，并将以几种编程语言提供参考实现，以便于采用。预计这些新的“压缩结构生物信息学”工具将使变革性研究成为可能，正如美国国立卫生研究院的“从大数据到知识”倡议所希望的那样。

 DESCRIPTION (provided by applicant): The Protein Data Bank (PDB) archive has doubled in size since 2008 and exceeded 100,000 entries in 2014. At the same time, the size and complexity of structures are increasing dramatically, for example the recently determined structure of the HIV-capsid contains about 2.5 million atoms. The emerging techniques of integrative Structural Biology are starting to determine structures of molecular machines in the mega-Dalton range by combining cryo-Electron Microscopy, Small-Angle X-ray Scattering, X-ray, and NMR at increasingly higher resolution. Interactive visualization of large complexes exceeds available network bandwidth and memory of typical scientists' desktops, laptops, or mobile devices. Large-scale structural analyses and queries of the archive have become a Big Data challenge. To make these structures accessible to all scientists, educators, and students, new ways of representing these data are required. In domains such as high-definition television, satellite communication, video or audio streaming, high-efficiency compression has been key to deliver interactive media to phones, tablets, laptops, and desktops. A similar trend has emerged in the handling of whole genome sequence data. An entire discipline "Compressive Genomics" has been developed to deal with data compression and development of algorithms to process these data. This proposal introduces the concept of "Compressive Structural Bioinformatics", a set of compression algorithms, applications, and workflows that analyze and visualize large structures and large sets of structures at an unprecedented speed (100-1000 fold speedup) and with minimal client side overhead. The aims of this project are: 1. Develop a compact and extensible representation of 3-D biomolecular structures, 2. Enable interactive visualization of large complexes by reducing network bandwidth and enabling data streaming, 3. Enable large-scale analyses of the PDB archive for I/O bound workflows, and 4. Develop open source software libraries. Through collaboration with developers of widely used visualization applications and distributed data-parallel workflow systems, the new techniques will be implemented, benchmarked, and reference implementations will be provided in several programming languages for easy adoption. It is expect that these new "Compressive Structural Bioinformatics" tools will enable transformative research as intended by the NIH's Big Data to Knowledge initiative.