Multi-Resolution Docking Methods for Electron Microscopy

 描述(由申请人提供)：在过去的十年中，通过电子显微镜对大分子组件进行3D成像以及开发将得到的体积图与原子分辨率结构相关联的计算算法方面取得了重大进展。拟议研究的总体目标是进一步开发电子显微镜(EM)的计算拟合和验证工具。我们打算建立新的建模、可视化和模拟技术，作为原子结构和EM密度之间的桥梁。拟议的多尺度软件将有助于常规确定生物分子组件的大规模结构，并有助于验证结构模型，这些模型将存入公共数据库，如蛋白质数据库(PDB)和EM数据库(EMB)。需要解决的关键问题包括：(I)如何改进、验证和传播用于中分辨率(8-15？)冷冻电子显微镜的成熟的匹配算法？(2)如何在低分辨率(4-5 nm)冷冻电子断层图像或树脂镶嵌样品块的聚焦离子束球磨中准确识别和分割亚细胞组件的几何特征？(3)鉴于直接探测相机最近实现的分辨率提高，如何系统地描述高分辨率(2-10？)密度模式，并根据数据中的本地签名验证原子模型？我们将采用一种新的建模范式来进行这些研究，即同时精化多个子单元。这种方法是基于“系统”的观点，因为生物集合在空间域中表现出“涌现行为”，即整体大于其各部分的总和。新的范式与对接协议相结合，提高了模型的准确性，并为上述三个领域的新的全局拟合应用打开了大门。此外，我们还将使用本地签名的统计分析和机器学习 以补充全球战略。这笔赠款支持的合作努力将包括改进细胞骨架细丝、分子马达、染色质纤维和毛细胞立体纤毛。算法和方法的发展将通过SUTUS和Sculptor程序包使用的基于互联网的既定机制免费分发。

 DESCRIPTION (provided by applicant): In the past decade, significant progress was made in 3D imaging of macromolecular assemblies via electron microscopy and in the development of computational algorithms that relate the resulting volumetric maps to atomic-resolution structures. The overall goal of the proposed research is to further develop computational fitting and validation tools for electron microscopy (EM). We intend to establish new modeling, visualization, and simulation techniques that would serve as bridges between atomic structures and EM densities. The proposed multi-scale software will aid in the routine determination of large-scale structures of biomolecular assemblies and in the validation of structural models that will be deposited to public databases such as the Protein Data Bank (PDB) and the EM Data Bank (EMDB). Key questions to be addressed include the following: (i) How can one improve, validate, and disseminate well-established matching algorithms for intermediate-resolution (8-15 Å) cryo-electron microscopy? (ii) How can one accurately identify and segment geometric features of subcellular assemblies in low-resolution (4-5 nm) cryo-electron tomograms or in focused ion beam milling of resin-embedded specimen blocks? (iii) Given the recent increase in resolution achieved with direct detection cameras, how can one systematically characterize high-resolution (2-10 Å) density patterns and validate atomic models based on local signatures in the data? We will adapt a new modeling paradigm for these studies, namely simultaneous refinement of multiple subunits. This approach is based on a "systems" perspective because biological assemblies exhibit "emergent behavior" in the spatial domain, that is, the whole is more than the sum of its parts. The new paradigm, in combination with docking protocols, improves model accuracy and opens the door to new global fitting applications in the above three areas. In addition, we will use statistical analysis and machine learning of local signatures to complement the global strategies. The collaborative efforts supported by this grant will include refinement of cytoskeletal filaments, molecular motors, chromatin fibers, and hair cell stereocilia. The algorithmic and methodological developments will be distributed freely through the established internet-based mechanisms used by the Situs and Sculptor packages.