DIALS / CCTBX: Serial crystallography computational methods aimed at biomolecular function

对人类健康至关重要的基本生化机制源于对大型生物分子（包括蛋白质和核酸）结构的理解。x射线晶体学已成为揭示其结构和功能的关键方法。该项目将开发能够使用连续x射线晶体学技术所需的计算方法。在第三代同步加速器光束线或x射线自由电子激光器（XFEL）上进行的连续晶体学正在成为一种确定分子结构的方法，这种方法使用短x射线脉冲探测一次晶体，然后交换新样品。这与传统的单晶实验不同，在传统的实验中，晶体在光束中旋转以组装完整的数据集，但这需要大晶体，再加上低温冷却以减缓辐射损伤的影响。相比之下，连续晶体学是用极短的x射线脉冲进行的，它在辐射损伤发生之前探测结构，并且在正常的生理温度下，可以显示所有可用的分子构象。软件工具DIALS（先进光源衍射集成）和CCTBX（计算晶体学工具箱）从由布拉格点组成的衍射图中提取信息，并对其进行分析，最终得到分子结构。这一建议重新检查了已经存在了几十年的既定数据处理模式，并倾向于为系列晶体学专门定制的新模型，对以前没有得到适当处理的测量进行系统修正。这将导致精度的提高，甚至达到定位蛋白质中单个电子的水平。软件将与全球多个XFEL光源合作部署，包括但不限于LCLS（斯坦福大学）、EuXFEL（德国）和SwissFEL（瑞士），以及SSRL（斯坦福大学）、ESRF（法国）和Diamond（英国）等多个同步加速器光源。代码将以开源、面向社区的软件体系结构发布，束线科学家可以对其进行调整，以适应新的仪器，在这个硬件快速发展预计将持续多年的领域。

Basic biochemical mechanisms fundamental to human health arise from understanding the structure of large biological molecules, both proteins and nucleic acids. X-ray crystallography has been a key method for uncovering their structure and function. This project will develop computational methods needed to enable the use of serial X-ray crystallography techniques. Serial crystallography, performed at either third generation synchrotron beamlines or X-ray free-electron lasers (XFEL), is emerging as a way to determine molecular structure using crystals that are probed once with a short X-ray pulse and then exchanged for a new sample. This a departure from traditional single-crystal experiments where the crystal is rotated in the beam to assemble a full data set, but which require large crystals, coupled with cryocooling to slow down the effects of radiation damage. Serial crystallography, in contrast, is performed with an extremely short X-ray pulse, which probes the structure before radiation damage occurs, and at normal physiological temperatures, where the full range of available molecular conformations can be revealed. The software toolkits DIALS (Diffraction Integration for Advanced Light Sources) and CCTBX (Computational Crystallography Toolbox) extract information from the diffraction pattern consisting of Bragg spots, the analysis of which eventually leads to molecular structure. This proposal re-examines the established data processing patterns that have existed for many decades, and favors new models that are specifically customized for serial crystallography, making systematic corrections to the measurements that have not previously been treated properly. This will lead to improved accuracy, even to the level of locating a single electron in a protein. Software will be deployed in cooperation with several XFEL lightsources worldwide including but not limited to LCLS (Stanford), EuXFEL (Germany), and SwissFEL (Switzerland), and at several synchrotron sources such as SSRL (Stanford), ESRF (France), and Diamond (UK). Code will be distributed in an open source, community-oriented software architecture that can be adapted by beamline scientists to accommodate new instrumentation, in a field where rapid hardware advances are expected to continue for many years.