An automated pipeline for macromolecular structure discovery in cellular electron cryo-tomography

细胞电子冷冻断层扫描中大分子结构发现的自动化流程

• 批准号: 9769773
• 负责人: NIELS VOLKMANN
• 金额: $ 92.84万
• 依托单位: SANFORD BURNHAM PREBYS MEDICAL DISCOVERY INSTITUTE
• 依托单位国家: 美国
• 财政年份: 2018
• 资助国家: 美国
• 起止时间: 2018-09-01 至 2020-07-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/9769773
• 关键词: Address; Algorithms; Artificial Intelligence; Big Data; Biological; Biology; Cancer Diagnostics; Cell physiology; Cells; Classification; Collection; Complex; Computing Methodologies; Cryo-electron tomography; Cryoelectron Microscopy; Data; Data Analyses; Data Collection; Data Set; Detection; Development; Disease; Electron Microscopy; Electrons; Environment; Floods; Goals; Health; Heart; Human; Image; Imaging Device; Individual; Knowledge; Laboratories; Methodology; Methods; Molecular; Molecular Structure; Morphology; Pattern; Pharmaceutical Preparations; Phase; Preparation; Process; Production; Real-Time Systems; Research Personnel; Resolution; Sampling; Scientist; Stimulus; Structure; System; Techniques; Technology; Tomogram; Validation; Variant; Visualization software; base; computer framework; convolutional neural network; deep learning; electron tomography; experience; imaging detection; improved; insight; knowledge base; learning strategy; nanomachine; novel diagnostics; particle; programs; reconstruction; response; software development; statistics; tomography; tool; virtual

SUMMARY – OVERALL Cellular cryo-tomography has emerged as a critical tool for the visualization and structural study of the molecular nanomachines at the heart of cellular function. Although the basic electron cryo-tomography technique has been used for several decades, the technology is being revolutionized by recent advances in sample preparation, electron cryo-microscopy hardware, improved capabilities for automatic data collection, direct electron detection imaging devices, and phase plate technologies. Combined, these advances led to the ability to generate extraordinarily large numbers of cellular cryo-tomograms of exquisite quality. In principle, such large data sets offer insights into cellular variation in disease states as well as better insights into basic cellular function, opening new possibilities for studying the underpinnings of health and disease at the finest possible level, potentially leading to completely new diagnostics for cancer and other cell-altering diseases. However, collection of cellular data is now at a far faster rate than can currently be analyzed with existing methods, producing a serious barrier to progress: to match the data production rates of a single laboratory, at least 50 experienced scientists would need to handle the data analysis. The primary goal of this Program Project is to establish quantitative and highly automated tools for the reconstruction and interpretation of highly complex cellular tomographic data. We have assembled a highly synergistic team of PIs with complimentary expertise in cutting-edge computational and experimental electron microscopy techniques to achieve this goal through collaborative efforts. Project 1 (Hanein & Penczek) focuses on development and implementation of tomogram quality assessment and validation techniques and on experimentally guided optimization of data collection strategies. Project 2 focuses on automatic tomographic reconstruction technology, extraction of various features from the tomograms, and the analysis of distribution patterns derived from the extracted features. Project 3 focuses on development of quantitative tools for tomogram annotation through deep learning and sub-tomogram alignment as well as interactive visualization tools. The set of highly automated tools developed in this Program Project will permit us to interpret 5–10x as much data as is possible using existing methods, greatly expanding the types of cellular variations we can effectively study.

摘要 - 总体 细胞冷冻术已成为可视化和结构研究的关键工具 分子纳米机器在细胞功能的核心。虽然基本电子冷冻术 技术已经使用了几十年了，该技术已被最近的进步彻底改变 样品制备，电子冷冻微镜硬件，提高自动数据收集功能， 直接电子检测成像设备和相板技术。结合了这些进步导致 能够产生大量的专有质量的蜂窝冷冻符号图。原则， 如此大的数据集提供了对疾病状态细胞变异的见解，以及对基本的更好的见解 细胞功能，为研究健康和疾病的基础开辟了新的可能性 可能的水平，有可能导致癌症和其他改变细胞的疾病的全新诊断。 但是，现在的蜂窝数据收集的速度要比现有的要快得多 方法，产生严重的进度障碍：与单个实验室的数据生产率相匹配，at 至少有50位经验丰富的科学家需要处理数据分析。 该计划项目的主要目标是建立定量和高度自动化的工具 重建和解释高度复杂的细胞层析成像数据。我们组装了一个高度 PI的协同团队具有免费的专业知识 显微镜技术通过协作努力实现这一目标。项目1（Hanein＆Penczek）重点 关于断层图质量评估和验证技术的开发和实施 实验指导的数据收集策略的优化。项目2专注于自动层析成像 重建技术，从断层图中提取各种特征以及分布分析 从提取的特征得出的模式。项目3的重点是开发定量工具 通过深度学习和子图像对齐方式以及交互式可视化的横向图注释 工具。该计划项目中开发的一系列高度自动化工具将使我们能够将5-10X解释为 尽可能使用现有方法的大量数据，大大扩展了我们可以的细胞变化类型 有效研究。