Centre for Structural Systems Biology (CSSB), Hamburg, Germany.
European Molecular Biology Laboratory Hamburg, Hamburg, Germany.
Nat Protoc. 2022 Jan;17(1):152-176. doi: 10.1038/s41596-021-00640-z. Epub 2021 Nov 29.
Integrative modeling enables structure determination of macromolecular complexes by combining data from multiple experimental sources such as X-ray crystallography, electron microscopy or cross-linking mass spectrometry. It is particularly useful for complexes not amenable to high-resolution electron microscopy-complexes that are flexible, heterogeneous or imaged in cells with cryo-electron tomography. We have recently developed an integrative modeling protocol that allowed us to model multi-megadalton complexes as large as the nuclear pore complex. Here, we describe the Assembline software package, which combines multiple programs and libraries with our own algorithms in a streamlined modeling pipeline. Assembline builds ensembles of models satisfying data from atomic structures or homology models, electron microscopy maps and other experimental data, and provides tools for their analysis. Compared with other methods, Assembline enables efficient sampling of conformational space through a multistep procedure, provides new modeling restraints and includes a unique configuration system for setting up the modeling project. Our protocol achieves exhaustive sampling in less than 100-1,000 CPU-hours even for complexes in the megadalton range. For larger complexes, resources available in institutional or public computer clusters are needed and sufficient to run the protocol. We also provide step-by-step instructions for preparing the input, running the core modeling steps and assessing modeling performance at any stage.
整合建模通过结合来自多个实验来源的数据，如 X 射线晶体学、电子显微镜或交联质谱，实现了对大分子复合物结构的测定。对于不适合高分辨率电子显微镜的复合物，如柔性、异质或在细胞中用冷冻电子断层扫描成像的复合物，它特别有用。我们最近开发了一种整合建模方案，使我们能够对核孔复合物等兆道尔顿大小的多聚体复合物进行建模。在这里，我们描述了 Assembline 软件包，它将多个程序和库与我们自己的算法结合在一个简化的建模管道中。Assembline 根据原子结构或同源模型、电子显微镜图谱和其他实验数据构建满足数据要求的模型集合，并提供用于分析这些模型的工具。与其他方法相比，Assembline 通过多步过程实现了对构象空间的有效采样，提供了新的建模约束，并包括一个独特的配置系统来设置建模项目。即使对于兆道尔顿范围内的复合物，我们的方案也能在不到 100-1000 个 CPU 小时内实现全面采样。对于更大的复合物，则需要机构或公共计算机集群中的资源，并足以运行该方案。我们还提供了逐步的说明，用于准备输入、运行核心建模步骤，并在任何阶段评估建模性能。
