其中一些工作是在纽约结构生物学中心的西蒙斯电子显微镜中心进行的，得到了西蒙斯基金会（SF349247）、美国国立卫生研究院（U24 GM129539）和纽约州议会的支持。

使用 Smart Leginon Autoscreen 加速大分子结构分析

<ol>
	<li>Noble, A. J., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Routine+single+particle+CryoEM+sample+and+grid+characterization+by+tomography.">Routine single particle CryoEM sample and grid characterization by tomography.</a> eLife. 7, e34257 (2018).</li><li>Weissenberger, G., Henderikx, R. J. M., Peters, P. J. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Understanding+the+invisible+hands+of+sample+preparation+for+cryo-EM.">Understanding the invisible hands of sample preparation for cryo-EM.</a> Nature Methods. 18 (5), 463-471 (2021).</li><li>Carragher, B., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Current+outcomes+when+optimizing+'standard'+sample+preparation+for+single-particle+cryo-EM.">Current outcomes when optimizing 'standard' sample preparation for single-particle cryo-EM.</a> Journal of Microscopy. 276 (1), 39-45 (2019).</li><li>Kampjut, D., Steiner, J., Sazanov, L. A. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Cryo-EM+grid+optimization+for+membrane+proteins.">Cryo-EM grid optimization for membrane proteins.</a> iScience. 24 (3), 102139 (2021).</li><li>Suloway, C., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Automated+molecular+microscopy:+The+new+Leginon+system.">Automated molecular microscopy: The new Leginon system.</a> Journal of Structural Biology. 151 (1), 41-60 (2005).</li><li>Mastronarde, D. N. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Automated+electron+microscope+tomography+using+robust+prediction+of+specimen+movements.">Automated electron microscope tomography using robust prediction of specimen movements.</a> Journal of Structural Biology. 152 (1), 36-51 (2005).</li><li>Koh, A., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Routine+collection+of+high-resolution+cryo-EM+datasets+using+200+KV+transmission+electron+microscope.">Routine collection of high-resolution cryo-EM datasets using 200 KV transmission electron microscope.</a> Journal of Visualized Experiments. 181, 63519 (2022).</li><li>Cheng, A., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Fully+automated+multi-grid+cryoEM+screening+using+Smart+Leginon.">Fully automated multi-grid cryoEM screening using Smart Leginon.</a> IUCrJ. 10 (1), 77-89 (2023).</li><li>Kim, P. T., Noble, A. J., Cheng, A., Bepler, T. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Learning+to+automate+cryo-electron+microscopy+data+collection+with+Ptolemy.">Learning to automate cryo-electron microscopy data collection with Ptolemy.</a> IUCrJ. 10 (1), 90-102 (2023).</li><li>Bepler, T., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Smart+data+collection+for+CryoEM.">Smart data collection for CryoEM.</a> Journal of Structural Biology. 214 (4), 107913 (2022).</li><li>Bouvette, J., Huang, Q., Riccio, A. A., Copeland, W. C., Bartesaghi, A., Borgnia, M. J. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Automated+systematic+evaluation+of+cryo-EM+specimens+with+SmartScope.">Automated systematic evaluation of cryo-EM specimens with SmartScope.</a> eLife. 11, e80047 (2022).</li><li>Deng, Y., Grollios, F., Kohr, H., van Knippenberg, B., Janus, M., Caglar, F. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Smart+EPU:+SPA+Getting+Intelligent.">Smart EPU: SPA Getting Intelligent.</a> Microscopy and Microanalysis. 27 (S1), 454-455 (2021).</li><li>Fan, Q., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=CryoRL:+Reinforcement+learning+enables+efficient+cryo-EM+data+collection.">CryoRL: Reinforcement learning enables efficient cryo-EM data collection.</a> arXiv. , (2022).</li><li>Li, Y., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Optimized+path+planning+surpasses+human+efficiency+in+cryo-EM+imaging.">Optimized path planning surpasses human efficiency in cryo-EM imaging.</a> bioRxiv. Biophysics. , (2022).</li><li>Lander, G. C., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Appion:+An+integrated,+database-driven+pipeline+to+facilitate+EM+image+processing.">Appion: An integrated, database-driven pipeline to facilitate EM image processing.</a> Journal of Structural Biology. 166 (1), 95-102 (2009).</li><li>Rice, W. J., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Routine+determination+of+ice+thickness+for+cryo-EM+grids.">Routine determination of ice thickness for cryo-EM grids.</a> Journal of Structural Biology. 204 (1), 38-44 (2018).</li><li>Russo, C. J. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=The+potential+for+seeing+molecules+in+cells,+CZ+Imaging+Institute's+Hardware+Frontiers+for+CryoET+Workshop.">The potential for seeing molecules in cells, CZ Imaging Institute's Hardware Frontiers for CryoET Workshop.</a> CZ Imaging Institute's Hardware Frontiers for CryoET Workshop. , (2023).</li><li>Punjani, A., Rubinstein, J. L., Fleet, D. J., Brubaker, M. A. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=cryoSPARC:+algorithms+for+rapid+unsupervised+cryo-EM+structure+determination.">cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination.</a> Nature Methods. 14 (3), 290-296 (2017).</li><li>Budell, W. C., Allegri, L., Dandey, V., Potter, C. S., Carragher, B. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Cryo-electron+microscopic+grid+preparation+for+time-resolved+studies+using+a+novel+robotic+system,+Spotiton.">Cryo-electron microscopic grid preparation for time-resolved studies using a novel robotic system, Spotiton.</a> Journal of Visualized Experiments. 168, 62271 (2021).</li><li>Darrow, M. C., Booth, T., Moore, J. P., Doering, K., Thaw, P., King, R. S. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Enabling+a+paradigm+shift+in+CryoEM+sample+preparation+with+chameleon.">Enabling a paradigm shift in CryoEM sample preparation with chameleon.</a> Microscopy and Microanalysis. 27 (S1), 524-525 (2021).</li></ol>

提交人声明，他们没有相互竞争的经济利益。

在此协议中，我们描述了 Smart Leginon Autoscreen 的管道，以及那些刚接触收集软件的人的基本 Leginon 用法。到 2024 年底，单颗粒冷冻电镜有望成为最高效的三维 （3D） 蛋白质结构分离技术17.单颗粒冷冻电镜管道由几个步骤组成，这些步骤不断优化以提高数据质量和通量。 图 2 显示了最常见的步骤（样品制备、网格制备、筛选时间和工作量、高分辨率收集时间、实时处理和完整的后处理）以及可以改进的管道的其他组件（筛选显微镜通道、载物台速度和精度、相机速度和高分辨率显微镜通道）。大多数步骤的结果都会成为先前步骤的反馈循环（ 图 2 中的蓝色箭头），使整个管道高度相互依赖。 图 2 中的每个步骤都涂有颜色，以近似该步骤相对于其他步骤的瓶颈程度。智能 Leginon Autoscreen 将筛选 12 个网格的时间和精力从 6 小时缩短到不到 10 分钟，从而缓解了瓶颈，并允许更快速地反馈样品/网格制备（图 3）。
《议定书》中有几个关键步骤，如 图1所示。至关重要的是，用于创建模板会话的网格必须代表要筛选的其余网格。重要的是，Leginon 会记住整个设置过程中的所有设置，以创建模板会话（ 图 1 中的蓝色步骤），这允许每次更快地设置循环模板会话。创建模板会话时，最关键的步骤是设置所有放大倍率的定位，以便参数和阈值反映要筛选的网格的预期变化。各种“测试”按钮可提高此设置过程的效率。在 Autoscreen 会话期间，监控 Appion 中的前几个网格以快速检测任何问题并尽快在 Leginon 内部修复它们至关重要。
SEMC 的典型工作流程是将 Autoscreen 数据输入到 CryoSPARC Live18 中，并使用这些附加信息将反馈回路告知样品/网格制备。在研究人员和操作员密集的冷冻电镜优化日期间，有关样品和网格条件的信息被反馈到样品和网格制备中，而Autoscreen仍在筛选网格。这允许每周冻结和筛选数十个网格8.
Smart Leginon Autoscreen 适用于在 SEMC 观察到的大多数 （80%-90%） 有孔的网格和条件。其余 10%-20% 的网格包括那些有时效果不佳的网格 - 孔和基板之间对比度差异最小的网格;具有较小孔和间距的网格（例如，0.6/0.8） - 以及通常不切实际地跨多个网格进行定位的网格 - Spotiton/Chameleon19,20 网格，由整个网格中的样本条纹组成;花边网格。请注意，带有 Autoscreen 的倾斜网格集合正在开发中，但尚未推出。可以修改协议以使用 Spotiton/Chameleon 网格，方法是首先手动成像条纹区域以确定窄参数阈值，然后在步骤 2.1.7.4 中尝试将更大和更小的方块分别组合在一起，然后从带有冰的组中选择目标。此修改的目标是让 Smart Leginon 将空方块和非空方块分成两组。如果找到参数，它们可能无法很好地扩展到要筛选的其余网格。也可以修改协议以使用花边网格，方法是删除步骤 2.1.9.1 中的hl_finding.sh脚本，并根据需要配置参数以针对较亮/较暗的区域。根据冰的厚度和网格材料，这种修改的成功率可能因网格而异。
在自动屏幕会话期间进行故障排除是可能的，有时是适当的。在自动采集过程中，可以更改预设（例如，散焦）和瞄准参数（例如，孔瞄准阈值）。在收集自动屏幕会话时，无法取消网格会话，因为它将终止 autoscreen.py。但是，“目标”节点中的“中止”按钮可用于跳过网格的任何部分或整个网格。有时，autoscreen.py 可能会使用过多的内存并冻结，提供两个选项：“强制退出”或“等待”。如果选择“强制退出”，则整个脚本将终止，要求用户重新运行脚本以应用于其余网格进行筛选。如果选择“等待”，脚本将继续，并且可以更改设置以防止将来冻结，例如，关闭“曝光”节点中的图像显示、减小图集中的像素大小或运行内存清除脚本。如果程序在未提供这两个选项的情况下冻结，则内存错误可能无法自行解决，从而导致采集暂停。在这种情况下，“强制退出”选项可能很有用。
SEMC经常使用Smart Leginon Autoscreen。随着单颗粒冷冻电镜管道中的瓶颈不断减少，冷冻电镜的采用将继续增加，以帮助回答生物学问题。该协议是朝着优化整个管道的方向迈出的一步，它为显著减少反馈循环提供了清晰的路径。

单颗粒冷冻电子显微镜 （cryoEM） 可以对纯化的大分子复合物进行近原子分辨率的结构测定。单个颗粒冷冻电镜实验只需要从具有不同样品和网格条件的更大网格中选择一个或两个精心选择的网格。检查这些网格的显微镜筛选需要以多个放大倍率对每个网格进行成像，以确定哪个网格满足高分辨率数据收集的最关键要求，包括冰厚、用于完整数据收集的足够区域、蛋白质纯度、蛋白质浓度、蛋白质稳定性和最小首选取向问题1.针对这些关键要求进行优化通常涉及显微镜筛选与制备条件（如蛋白质生产、缓冲液选择、潜在去污剂和网格类型 2、3、4 ）之间的反馈（图 1）。传统的网格筛选是使用 Leginon5、SerialEM6 和 EPU7 等软件手动或半手动执行的。传统筛选要求显微镜操作员在显微镜前花费数小时来筛选多个网格，这在高分辨率单颗粒工作流程中造成了重大瓶颈，因为操作员需要死记硬背的操作而不是样品/网格优化。
此前，已经介绍了Smart Leginon Autoscreen和底层机器学习软件Ptolemy，并介绍了它们的底层方法和算法以及示例8,9。其他几个软件包能够或致力于全自动多网格筛选10，包括SmartScope11、Smart EPU12和CryoRL13,14。为了解决筛选瓶颈，Smart Legino允许用户首先在模板显微镜会话中设置筛选参数，然后使用该模板会话的参数作为模板，在显微镜自动加载器中筛选完整的网格盒。消除了暗盒筛选过程中的所有手动工作，从而使优化反馈回路能够更高效地进行。
在该协议中，描述了完整的Smart Leginon Autoscreen工作流程，以便阅读器可以独立执行全自动多网格冷冻电镜筛选。对于那些刚接触 Leginon 的人来说，协议的第一部分描述了传统的 Leginon 用法。这些知识由在几台自动加载显微镜上的多年经验组成，然后在随后的协议的Smart Legino部分中建立起来。其他教程视频可以在 https://memc.nysbc.org 上找到。

<table border="1" fo:keep-together.within-page="1" fo:keep-with-next.within-page="always">
	<tbody>
		<tr>
			<td>Glacios cryoTEM&#160;</td>
			<td>Thermo Fisher Scientific&#160;</td>
			<td>GLACIOSTEM</td>
			<td>FEG, 200 keV, Falcon 3EC camera&#160;</td>
		</tr>
		<tr>
			<td>Krios cryoTEM&#160;</td>
			<td>Thermo Fisher Scientific&#160;</td>
			<td>KRIOSG4TEM</td>
			<td>XFEG, 300 keV,&#160;Gatan BioQuantum energy filter, Gatan K3 camera&#160;</td>
		</tr>
		<tr>
			<td>Leginon&#160;</td>
			<td>Simons Electron Microscopy Center&#160;</td>
			<td></td>
			<td>http://leginon.org&#160;</td>
		</tr>
		<tr>
			<td>Ptolemy&#160;</td>
			<td>Simons Machine Learning Center&#160;</td>
			<td></td>
			<td>https://github.com/SMLC-NYSBC/ptolemy&#160;</td>
		</tr>
	</tbody>
</table>

cryo-electron microscopy screening automation across multiple grids using smart leginon

在过去十年中，冷冻电子显微镜 （cryoEM） 技术的进步使结构生物学家能够常规地将大分子蛋白质复合物解析到接近原子的分辨率。整个冷冻电镜管道的一般工作流程包括在进行高分辨率数据收集之前，在样品制备、冷冻电镜网格制备和样品/网格筛选之间进行迭代。样品/网格制备和筛选之间的迭代通常是研究人员的主要瓶颈，因为每个迭代实验都必须针对样品浓度、缓冲条件、网格材料、网格孔大小、冰厚度和冰中的蛋白质颗粒行为等变量进行优化。此外，一旦这些变量得到令人满意的确定，在相同条件下制备的网格在是否准备好进行数据收集方面差异很大，因此建议在选择用于高分辨率数据收集的最佳网格之前进行额外的筛选。这种样品/网格制备和筛选过程通常会消耗数十个网格和数天的操作人员时间。此外，筛选过程仅限于操作人员/显微镜的可用性和显微镜的可及性。在这里，我们演示了如何使用 Leginon 和 Smart Leginon Autoscreen 来自动化大多数冷冻电镜网格筛选。Autoscreen 结合了机器学习、计算机视觉算法和显微镜处理算法，无需操作员不断手动输入。Autoscreen可以使用自动样本交换盒系统自主加载和成像具有多尺度成像的网格，从而实现整个盒的无人值守网格筛选。因此，与使用以前的方法相比，使用 Autoscreen 筛选 12 个网格的操作员时间可以减少到 ~10 分钟，而使用以前的方法则需要 ~6 小时，这些方法无法解释网格之间的高可变性。该协议首先介绍基本的 Leginon 设置和功能，然后逐步演示从创建模板会话到 12 格自动筛选会话结束的自动筛选功能。

Single particle cryo-electron microscopy (cryoEM) allows for near-atomic resolution structure determination of purified macromolecular complexes. A single particle cryoEM experiment only requires one or two well-chosen grids selected from a much larger set of grids with varying sample and grid conditions. Microscope screening to examine these grids entails imaging each grid at several magnifications to determine which grid satisfies most key requirements for high-resolution data collection, including ice thickness, sufficient areas for full data collection, protein purity, protein concentration, protein stability, and minimal preferred orientation issues1. Optimizing for these key requirements often involves feedback between screening at the microscope and preparation conditions such as protein production, buffer selection, potential detergents, and grid type2,3,4 (Figure 1). Conventional grid screening is performed manually or semi-manually with software such as Leginon5, SerialEM6, and EPU7. Conventional screening requires the microscope operator to spend hours at the microscope to screen several grids, which creates a significant bottleneck in the high-resolution single particle workflow by occupying the operator with rote operations rather than sample/grid optimization.
Previously, Smart Leginon Autoscreen and the underlying machine learning software, Ptolemy, have been introduced, and their underlying methods and algorithms together with examples have been described8,9. Several other software packages are either capable of or working towards fully automated multi-grid screening10, including SmartScope11, Smart EPU12, and CryoRL13,14. To address the screening bottleneck, Smart Leginon allows the user to first set up screening parameters in a template microscope session, then use that template session&#39;s parameters as a template to screen the full cassette of grids in the microscope autoloader. All manual work during the cassette screening is eliminated, which allows for the optimization feedback loop to proceed significantly more efficiently.
In this protocol, the full Smart Leginon Autoscreen workflow is described so that the reader may perform fully automated multi-grid cryoEM screening independently. For those new to Leginon, the first section in the protocol describes conventional Leginon usage. This knowledge is composed of several years of experience across several autoloader microscopes, which is then built upon in the subsequent Smart Leginon section of the protocol. Additional tutorial videos may be found on https://memc.nysbc.org.

作者聚焦：通过自动数据收集和分析技术增强冷冻电子显微镜

使用Smart Legino的冷冻电子显微镜跨多个网格进行自动化筛选

At the Simons Electron Microscopy Center, we aim to provide superb electron microscopy services, expertise, and training to local and national researchers, and to develop innovative methods to optimize workflows. Our scope covers service, development, machine learning, and biomedical research. We are trying to answer, how can we better optimize our services?Machine learning is changing the field of electron microscopy. Data collection is becoming automated. Data analysis tools are rapidly becoming more efficient.This includes denoising images, particle picking, 3G flexibility analysis, and model building. We regularly update software to account for different situations, such as different grid holes, tilted collection, or bugs identified through beta testing software in our production environment. We also integrate feedback from our users, so we are continuously providing features that are useful for them.Smart Leginon addresses the labor-intensive and time-consuming tasks of screening one or more grids on a cryoTEM. It enables operators and researchers to more quickly screen and analyze grids, while freeing up time to optimize grid and sample conditions in parallel. This also reduces microscope time and cost.Smart Leginon autoscreen offers fully unattended multigrid screening, which takes about 10 minutes to set up screening for 12 grids. Compared to manual screening, our protocol saves several hours per day. Since Smart Leginon's introduction, other free and commercial software has been developed for similar automation.

按照该协议，冷冻电镜筛选过程可以自动成功运行，适用于大多数 （80%-90%） 有孔的网格和条件。之前已经提出了几个示例和实验 8,9 以证明成功的 Smart Leginon Autoscreen 会话的预期结果。成功的自动筛选会话从 ~10 分钟的设置开始，通常会导致在大约 6 小时（每个网格 30 分钟）后自动筛选 12 个网格的完整盒，其中 3-5 个不同大小的正方形和每个正方形 3-5 个孔在高放大倍率下成像，使用户能够快速确定每个网格上样品的特征，并快速迭代样品/网格条件（图 3).有时，会话不成功，通常是由于自动屏幕定位破碎的方块，没有正确解释整个网格或整个方块上的大冰厚梯度，或者未能正确识别碳网格上的孔。此外，潜在的内存泄漏可能会导致 Leginon 因内存使用过多而崩溃，这可以通过释放 RAM 或重新启动计算机来解决，或者通过向计算机添加更多 RAM 来改善。
<img alt="Figure 1" class="xfigimg" src="/files/ftp_upload/66007/66007fig01.jpg"> 图 1：Smart Leginon Autoscreen 工作流程。Smart Leginon Autoscreen 工作流程的高级概述。首先，通过在要筛选的网格批次中为具有代表性的网格选择参数来创建模板会话。设置 Leginon 和创建模板会话只需不到 45 分钟。其次，将 Autoscreen 设置为使用模板会话参数来筛选盒中的所有网格。设置自动屏幕只需不到 10 分钟。最后，Autoscreen 结束筛选会话。<a href="https://www.jove.com/files/ftp_upload/66007/66007fig01large.jpg" target="_blank">请点击这里查看此图的较大版本.</a>
<img alt="Figure 2" class="xfigimg" src="/files/ftp_upload/66007/66007fig02.jpg"> 图 2： 自动筛选前的常规单颗粒冷冻电镜管道。在自动筛选之前，传统单颗粒冷冻电镜管道中最常见的步骤，以及可以改进的组件。每个步骤都涂有颜色，以近似该步骤相对于其他步骤的瓶颈程度。蓝色圆形箭头表示大多数步骤之间的几个反馈循环。几个步骤的通量很大程度上取决于样品、资金和研究人员的位置。 <a href="https://www.jove.com/files/ftp_upload/66007/66007fig02large.jpg" target="_blank">请点击这里查看此图的较大版本.</a>
<img alt="Figure 3" class="xfigimg" src="/files/ftp_upload/66007/66007fig03.jpg"> 图 3：具有代表性的 Smart Leginon Autoscreen 结果。在带有 BioQuantum 能量过滤器和 K3 相机的 TFS Krios 冷冻透射电镜上收集的遵循 Smart Leginon Autoscreen 协议的代表性多尺度图像。（A） 显示冷冻电镜网格概述的复合“图谱”图像。（B-F）来自网格图集中指定位置的多比例图像。自动选择第一行的低倍率图像、第二行的中倍率图像和第三行的高倍率图像，从薄到厚的冰块获取样品信息。Leginon 估计的冰厚显示在底部。比例尺在 （A） 中为 500 μm，第一行为 10 μm，第二行为 5 μm，第三行为 （B-F） 为 100 nm。此图经 Cheng et al.8 许可修改。<a href="https://www.jove.com/files/ftp_upload/66007/66007fig03large.jpg" target="_blank">请点击这里查看此图的较大版本.</a>
<table border="1" fo:keep-together.within-page="1" fo:keep-with-next.within-page="always"><tbody><tr><td></td>
			<td>gr：网格</td>
			<td>sq：正方形</td>
			<td>hln：孔</td>
			<td>风扇：自动对焦</td>
			<td>fcn：中心焦点</td>
			<td>enn： 曝光 </td>
		</tr><tr><td>放大</td>
			<td>210</td>
			<td>2600</td>
			<td>6700</td>
			<td>120000</td>
			<td>120000</td>
			<td>120000</td>
		</tr><tr><td>离 焦</td>
			<td>-0.0002</td>
			<td>-0.00015</td>
			<td>-0.00015</td>
			<td>-2 x 10-06 </td>
			<td>-7 x 10-07 </td>
			<td>-2.5 x 10-06 </td>
		</tr><tr><td>光斑尺寸</td>
			<td>5</td>
			<td>5</td>
			<td>4</td>
			<td>2</td>
			<td>2</td>
			<td>2</td>
		</tr><tr><td>强度</td>
			<td>1.1</td>
			<td>0.83</td>
			<td>0.65</td>
			<td>0.44</td>
			<td>0.44</td>
			<td>0.45</td>
		</tr><tr><td>尺寸 </td>
			<td>约1024 x 1024</td>
			<td>约1024 x 1024</td>
			<td>约1024 x 1024</td>
			<td>约1024 x 1024</td>
			<td>约1024 x 1024</td>
			<td>4096 x 4096</td>
		</tr><tr><td>抵消</td>
			<td>0, 0</td>
			<td>0, 0</td>
			<td>0, 0</td>
			<td>0, 0</td>
			<td>512, 512</td>
			<td>0, 0</td>
		</tr><tr><td>像素合并</td>
			<td>4 x 4</td>
			<td>4 x 4</td>
			<td>4 x 4</td>
			<td>4 x 4</td>
			<td>2 x 2</td>
			<td>1 x 1</td>
		</tr><tr><td>曝光时间（ms）</td>
			<td>200</td>
			<td>500</td>
			<td>500</td>
			<td>500</td>
			<td>500</td>
			<td>1000</td>
		</tr><tr><td>预曝光</td>
			<td>0</td>
			<td>0</td>
			<td>0</td>
			<td>0</td>
			<td>0</td>
			<td>0</td>
		</tr><tr><td>剂量 （e/Å2）</td>
			<td>--</td>
			<td>--</td>
			<td>--</td>
			<td>36.5</td>
			<td>36.5</td>
			<td>64.7</td>
		</tr><tr><td>保存原始帧</td>
			<td>不</td>
			<td>不</td>
			<td>不</td>
			<td>不</td>
			<td>不</td>
			<td>是的</td>
		</tr></tbody></table>表 1：西蒙斯电子显微镜中心 （SEMC） 使用 Glacios cryoTEM 和 Falcon 3EC 相机进行冷冻电镜网格筛选的预设参数。 显示了SEMC的Falcon 3EC相机在Glacios cryoTEM上常用的每个预设参数。不同的显微镜将具有不同的放大倍率，不同的实验将使用不同的参数，例如散焦和曝光时间。
<table border="1" fo:keep-together.within-page="1" fo:keep-with-next.within-page="always"><tbody><tr><td></td>
			<td>gr：网格</td>
			<td>sq：正方形</td>
			<td>hln：孔</td>
			<td>风扇：自动对焦</td>
			<td>fcn：中心焦点</td>
			<td>enn： 曝光 </td>
		</tr><tr><td>放大</td>
			<td>64</td>
			<td>1700</td>
			<td>2850</td>
			<td>75000</td>
			<td>75000</td>
			<td>75000</td>
		</tr><tr><td>离 焦</td>
			<td>0</td>
			<td>-5 x 10-05 </td>
			<td>-5 x 10-05 </td>
			<td>-1 x 10-06 </td>
			<td>-7 x 10-07 </td>
			<td>-2 x 10-06 </td>
		</tr><tr><td>光斑尺寸</td>
			<td>6</td>
			<td>9</td>
			<td>9</td>
			<td>6</td>
			<td>6</td>
			<td>7</td>
		</tr><tr><td>强度</td>
			<td>0.001</td>
			<td>约1.65 x 10-05 </td>
			<td>约1.5 x 10-05 </td>
			<td>4.3 x 10-07 </td>
			<td>4.3 x 10-07 </td>
			<td>约5.5×10-07 </td>
		</tr><tr><td>能量过滤器宽度</td>
			<td>--</td>
			<td>--</td>
			<td>--</td>
			<td>20</td>
			<td>20</td>
			<td>20</td>
		</tr><tr><td>尺寸</td>
			<td>约1024 x 1024</td>
			<td>约1024 x 1024</td>
			<td>约1024 x 1024</td>
			<td>约1024 x 1024</td>
			<td>约2048 x 2048</td>
			<td>4096 x 4096</td>
		</tr><tr><td>抵消</td>
			<td>0, 0</td>
			<td>0, 0</td>
			<td>0, 0</td>
			<td>0, 0</td>
			<td>0, 0</td>
			<td>0, 0</td>
		</tr><tr><td>像素合并 </td>
			<td>4 x 4</td>
			<td>4 x 4</td>
			<td>4 x 4</td>
			<td>4 x 4</td>
			<td>2 x 2</td>
			<td>1 x 1</td>
		</tr><tr><td>曝光时间（ms）</td>
			<td>500</td>
			<td>2000</td>
			<td>1000</td>
			<td>500</td>
			<td>300</td>
			<td>8700</td>
		</tr><tr><td>预曝光</td>
			<td>0</td>
			<td>0</td>
			<td>0</td>
			<td>0</td>
			<td>0</td>
			<td>0</td>
		</tr><tr><td>剂量 （e/Å2）</td>
			<td>--</td>
			<td>--</td>
			<td>--</td>
			<td>--</td>
			<td>--</td>
			<td>47.4</td>
		</tr><tr><td>保存原始帧</td>
			<td>不</td>
			<td>不</td>
			<td>不</td>
			<td>不</td>
			<td>不</td>
			<td>是的</td>
		</tr></tbody></table>表 2：使用 Krios cryoTEM 和 Selectris X 和 Falcon 4i 相机在 SEMC 进行冷冻电镜网格筛选的预设参数。 显示了带有 Selectris X 能量过滤器的 Krios 和 SEMC 的 Falcon 4i 相机上常用的每个预设的参数。不同的显微镜将具有不同的放大倍率，不同的实验将使用不同的参数，例如散焦和曝光时间。
<table border="1" fo:keep-together.within-page="1" fo:keep-with-next.within-page="always"><tbody><tr><td></td>
			<td>gr：网格</td>
			<td>sq：正方形</td>
			<td>hln：孔</td>
			<td>风扇：自动对焦</td>
			<td>fcn：中心焦点</td>
			<td>enn： 曝光 </td>
		</tr><tr><td>放大</td>
			<td>1550</td>
			<td>940</td>
			<td>2250</td>
			<td>81000</td>
			<td>81000</td>
			<td>81000</td>
		</tr><tr><td>离 焦</td>
			<td>0</td>
			<td>-5 x 10-05 </td>
			<td>-5 x 10-05 </td>
			<td>-1 x 10-06 </td>
			<td>-7 x 10-07 </td>
			<td>-2 x 10-06 </td>
		</tr><tr><td>光斑尺寸</td>
			<td>4</td>
			<td>8</td>
			<td>7</td>
			<td>6</td>
			<td>6</td>
			<td>6</td>
		</tr><tr><td>强度</td>
			<td>0.0015</td>
			<td>0.00017</td>
			<td>约7.3×10-05 </td>
			<td>约1.3 x 10-06 </td>
			<td>约1.3 x 10-06 </td>
			<td>约9.2 x 10-07 </td>
		</tr><tr><td>能量过滤器宽度</td>
			<td>--</td>
			<td>--</td>
			<td>50</td>
			<td>20</td>
			<td>20</td>
			<td>20</td>
		</tr><tr><td>尺寸</td>
			<td>约1024 x 1024</td>
			<td>约1440 x 1024</td>
			<td>约1440 x 1024</td>
			<td>约1440 x 1024</td>
			<td>约1008 x 1008</td>
			<td>5760 x 4092</td>
		</tr><tr><td>抵消 </td>
			<td>0, 0</td>
			<td>0, 0</td>
			<td>0, 0</td>
			<td>0, 0</td>
			<td>936, 519</td>
			<td>0, 0</td>
		</tr><tr><td>像素合并</td>
			<td>4 x 4</td>
			<td>8 x 8</td>
			<td>8 x 8</td>
			<td>8 x 8</td>
			<td>4 x 4</td>
			<td>2 x 2</td>
		</tr><tr><td>曝光时间（ms）</td>
			<td>250</td>
			<td>600</td>
			<td>600</td>
			<td>500</td>
			<td>500</td>
			<td>2100</td>
		</tr><tr><td>预曝光</td>
			<td>0</td>
			<td>0</td>
			<td>0</td>
			<td>0</td>
			<td>0</td>
			<td>0</td>
		</tr><tr><td>剂量 （e/Å2）</td>
			<td>--</td>
			<td>--</td>
			<td>--</td>
			<td>--</td>
			<td>--</td>
			<td>51</td>
		</tr><tr><td>保存原始帧</td>
			<td>不</td>
			<td>不</td>
			<td>不</td>
			<td>不</td>
			<td>不</td>
			<td>是的</td>
		</tr></tbody></table>表 3：使用 Krios 冷冻透射电镜和 BioQuantum 和 K3 相机在 SEMC 进行冷冻电镜网格筛选的预设参数。 显示了 SEMC 上带有 BioQuantum 能量过滤器和 K3 相机的 Krios 上常用的每个预设的参数。不同的显微镜将具有不同的放大倍率，不同的实验将使用不同的参数，例如散焦和曝光时间。
补充图 1：Smart Leginon 的方形定位设置和方形设置。 （A） 方形定位设置。（B） 方形设置。 <a href="https://www.jove.com/files/ftp_upload/66007/Supplementary%20Figure%20(1).pdf" target="_blank">请点击此处下载此文件。</a>
补充图 2：Smart Leginon 的 Hole Targeting 设置和 Hole 设置。 （A） 孔瞄准设置。（B） 孔设置。 <a href="https://www.jove.com/files/ftp_upload/66007/Supplementary%20Figure%20(2).pdf" target="_blank">请点击此处下载此文件。</a>
补充图 3：Smart Leginon 的 Exposure Targeting 设置和 Exposure 设置。 （A） 曝光定位设置。（B） 曝光设置。 <a href="https://www.jove.com/files/ftp_upload/66007/Supplementary%20Figure%20(3).pdf" target="_blank">请点击此处下载此文件。</a>
补充图 4：Smart Leginon 的对焦设置和对焦序列设置。 （A） 对焦设置。（B） 对焦序列设置 （Defocus1）。（C） 对焦序列设置 （Defocus2）。 <a href="https://www.jove.com/files/ftp_upload/66007/Supplementary%20Figure%20(4).pdf" target="_blank">请点击此处下载此文件。</a>
补充图 5：Smart Leginon 的Z_Focus设置和Z_Focus序列设置。 （A） Z_Focus设置。（B） Z_Focus 序列设置 （Stage_Tilt_Rough）。（C） Z_Focus 序列设置 （Stage_Tilt_Fine）。 <a href="https://www.jove.com/files/ftp_upload/66007/Supplementary%20Figure%20(5).pdf" target="_blank">请点击此处下载此文件。</a>
补充图 6：设置 Smart Leginon Square_Targeting参数后的示例图集。 蓝色圆圈是斑点，绿色加号是采集位置，棕色的“x”是当前阶段位置。 <a href="https://www.jove.com/files/ftp_upload/66007/Supplementary%20Figure%20(6).pdf" target="_blank">请点击此处下载此文件。</a>
补充图 7：设置 Smart Leginon Hole_Targeting参数后的示例图集。 紫色加号是格子位置，带方框的绿色加号是采集位置，蓝色加号是焦点位置。 <a href="https://www.jove.com/files/ftp_upload/66007/Supplementary%20Figure%20(7).pdf" target="_blank">请点击此处下载此文件。</a>
补充图 8：设置 Smart Leginon Exposure_Targeting参数后的示例图集。 蓝色圆圈是斑点，绿色加号是采集位置，蓝色加号是焦点位置。 <a href="https://www.jove.com/files/ftp_upload/66007/Supplementary%20Figure%20(8).pdf" target="_blank">请点击此处下载此文件。</a>
补充图 9：Smart Leginon Autoscreen 终端设置。<a href="https://www.jove.com/files/ftp_upload/66007/Supplementary%20Figure%20(9).pdf" target="_blank">请点击此处下载此文件。</a>
补充图 10：Smart Leginon Autoscreen GUI 设置。<a href="https://www.jove.com/files/ftp_upload/66007/Supplementary%20Figure%20(10).pdf" target="_blank">请点击此处下载此文件。</a>

Watch this Scientific Journal Video about Author Spotlight: Enhancing Cryo-Electron Microscopy by Automated Data Collection and Analysis Techniques at JoVE.com

冷冻电子显微镜 （cryoEM） 多网格筛选通常是一个繁琐的过程，需要数小时的注意力。该协议显示了如何设置标准 Leginon 集合和 Smart Leginon Autoscreen 以自动执行此过程。该协议可应用于大多数冷冻电镜多孔箔网格。

Advancements in cryo-electron microscopy (cryoEM) techniques over the past decade have allowed structural biologists to routinely resolve macromolecular ...

在西蒙斯电子显微镜中心，我们的目标是为当地和国家研究人员提供一流的电子显微镜服务、专业知识和培训，并开发创新方法来优化工作流程。我们的业务范围涵盖服务、开发、机器学习和生物医学研究。我们试图回答，我们如何才能更好地优化我们的服务？机器学习正在改变电子显微镜领域。数据收集正在变得自动化。数据分析工具正在迅速变得更加高效。这包括图像去噪、粒子拾取、3G 灵活性分析和模型构建。我们会定期更新软件以应对不同的情况，例如不同的网格孔、倾斜的收集或通过生产环境中的 beta 测试软件识别的错误。我们还整合了用户的反馈，因此我们不断提供对他们有用的功能。Smart Leginon 解决了在冷冻透射电镜上筛选一个或多个网格的劳动密集型和耗时任务。它使操作员和研究人员能够更快地筛选和分析网格，同时腾出时间并行优化网格和样品条件。这也减少了显微镜的时间和成本。Smart Leginon 自动屏幕提供完全无人值守的多网格筛选，设置 12 个网格的筛选大约需要 10 分钟。与人工筛查相比，我们的方案每天可节省数小时。自 Smart Leginon 推出以来，已经为类似的自动化开发了其他免费和商业软件。

cryo-electron-microscopy-screening-automation-across-multiple-grids

Research

JoVE Journal

Biochemistry

Cryo-Electron Microscopy Screening Automation Across Multiple Grids Using Smart Leginon

738 次观看.  Simons Electron Microscopy Center, New York Structural Biology Center. 在西蒙斯电子显微镜中心，我们的目标是为当地和国家研究人员提供一流的电子显微镜服务、专业知识和培训，并开发创新方法来优化工作流程。我们的业务范围涵盖服务、开发、机器学习和生物医学研究。我们试图回答，我们如何才能更好地优化我们的服务？机器学习正在改变电子显微镜领域。数据收集正在变得自动化。数据分析工具正在迅速变得更加高效。这...

视频: 作者聚焦：通过自动数据收集和分析技术增强冷冻电子显微镜

Advancements in cryo-electron microscopy (cryoEM) techniques over the past decade have allowed structural biologists to routinely resolve macromolecular protein complexes to near-atomic resolution. The general workflow of the entire cryoEM pipeline involves iterating between sample preparation, cryoEM grid preparation, and sample/grid screening before moving on to high-resolution data collection. Iterating between sample/grid preparation and screening is typically a major bottleneck for researchers, as every iterative experiment must optimize for sample concentration, buffer conditions, grid material, grid hole size, ice thickness, and protein particle behavior in the ice, amongst other variables. Furthermore, once these variables are satisfactorily determined, grids prepared under identical conditions vary widely in whether they are ready for data collection, so additional screening sessions prior to selecting optimal grids for high-resolution data collection are recommended. This sample/grid preparation and screening process often consumes several dozen grids and days of operator time at the microscope. Furthermore, the screening process is limited to operator/microscope availability and microscope accessibility. Here, we demonstrate how to use Leginon and Smart Leginon Autoscreen to automate the majority of cryoEM grid screening. Autoscreen combines machine learning, computer vision algorithms, and microscope-handling algorithms to remove the need for constant manual operator input. Autoscreen can autonomously load and image grids with multi-scale imaging using an automated specimen-exchange cassette system, resulting in unattended grid screening for an entire cassette. As a result, operator time for screening 12 grids may be reduced to ~10 min with Autoscreen compared to ~6 h using previous methods which are hampered by their inability to account for high variability between grids. This protocol first introduces basic Leginon setup and functionality, then demonstrates Autoscreen functionality step-by-step from the creation of a template session to the end of a 12-grid automated screening session.

Cryo-electron microscopy (cryoEM) multi-grid screening is often a tedious process that demands hours of attention. This protocol shows how to set up a standard Leginon collection and Smart Leginon Autoscreen to automate this process. This protocol can be applied to the majority of cryoEM holey foil grids.

科研

生物化学

Using Smart Leginon Autoscreen for CryoEM Grid Screening

To begin, set up the legend on interface, then navigate to application and click run. In the run application window, choose the Ptolemy application. Set main to the Legendon computer in scope and camera to the respective computers.Now import presets in the preset manager. Within the square targeting node settings, ensure that both sort targets by shortest path and enable auto targeting are selected. In the square node settings, confirm that the wait for node to process the image is checked.Add the square preset to the list if it's not already included. Under advanced settings, check set these apertures while imaging and ensure the aperture values are correct. To adjust whole targeting node settings, check allow for user verification of selected targets.Uncheck both queue up targets and skip automated hole finder. In the whole node setting, check wait for a node to process the image and the whole preset is in the list on the right. In advanced settings, check set these apertures while imaging and confirm the aperture values.In the exposure targeting node settings, select allow for user verification of selected targets. Ensure the Q up targets and skip automated hole finder are unchecked. To configure the exposure node settings, ensure the wait for a node to process the image is unchecked.Confirm that the exposure preset is included in the list on the right. In advanced settings, select set these apertures while imaging and verify that these aperture values are accurate. In the focus node settings, ensure wait for a node to process the image is unchecked.Verify that the auto focusing preset is listed on the right and set the desired autofocus accuracy to four times 10 to the negative six meters. In the focus node focus sequence, activate only two beam tilt auto focusing steps. To adjust the Z focus node settings ensure that wait for a node to process the image is unchecked.Confirm the whole preset is listed on the right and set the desired auto-focus accuracy to five times 10 to the negative five meters. In the Z focus node focus sequence, enable only two low magnification stage tilt steps. After determining the grid Z height and collecting an atlas, use the square targeting node in Ptolemy to locate squares.Observe the blue circles on each square. Hover over each blob to view their size as computed by Ptolemy and record the sizes of the largest and smallest blobs. To modify the threshold settings, adjust the minimum and maximum filter range to include desirable squares and exclude undesirable ones.Press the fine squares button on the top toolbar. Fine tune the filter range until the fine squares feature effectively targets the squares. In the acquisition settings, select appropriate values for the maximum number of targets and number of target group to sample.Once the parameters are satisfactory, click the play button to proceed. Then utilize the ruler tool in the whole targeting node to measure the diameter of a hole. In the template settings, ensure the measured diameter into final template diameter and click test.Adjust the diameter value until all holes display bright white peaks at their centers. Click test in the threshold settings. Modify the A value until the binarized image distinctly shows white areas solely where holes are present.In the blobs settings, measure a minimum distance from the edge with the ruler and input that value to exclude border targets. Filter blobs based on size, roundness and desired quantity. Hover over blobs to view their values and click test to review the current settings.Enter the radius of holes and the spacing between them in the lattice settings. Then click the 42 button to gauge the reference intensity value of a vacuum area. In the acquisition settings, check use subset of the acquisition targets and set the sample maximal value to a low number such as two.Define a broad range of ice thickness means in standard deviations. Use test targeting to randomize target selection based on the above values. Click the play button to proceed once all settings are satisfactory.In the whole settings, set the shell script to the hlfinding. sh script path in the Ptolemy installation. Set the minimum score to less than zero.Input the radius of the holes then click the show value button to measure the reference intensity value of a vacuum area. Click test to find the lattice of holes. In acquisition settings, check use subset of the acquisition targets and set the sample maximal value to a small number such as four.Set a wide range of ice thickness means and standard deviations. Click the play button when satisfied with all settings. Legendon will perform centric focus and collect high magnification images which can be seen in the exposure node.Check the next exposure targeting image to see if the settings are sufficient. Once satisfied, uncheck allow for user verification of selected targets in the exposure targeting and whole targeting settings. Once the grid is done screening, click file followed by exit to close Legendon.To set up smart Legendon auto screen, execute smart Legendon's autoscreen. py in a terminal window. Select GY and input a comma-separated list of grid slots to the screen.Then input full for the workflow. Input the template session name for new sessions and the template sessions Z height value. A GY will open to input the session name and select project associations for each grid.Leave the microscope unattended or monitor it in Legendon, Appian and the microscope computer.