Date Published: June 01, 2017
Publisher: International Union of Crystallography
Author(s): Rafael Fernandez-Leiro, Sjors H. W. Scheres.
The formal description of a workflow to cryo-EM structure determination in the RELION program allows standardization of procedures and on-the-fly image processing during data acquisition.
Recent advances in cryo-EM instrumentation and image-processing software have substantially expanded the scope of structure determination by single-particle analysis (Fernandez-Leiro & Scheres, 2016 ▸). As a result, the field is growing rapidly. With many new users turning to cryo-EM structure determination, efforts in methods development are increasingly focused on improving the accessibility of the technique. Examples of this are the development of robots for sample preparation and transfer of the sample into the microscope (Cheng et al., 2007 ▸; Vos et al., 2008 ▸; Kim et al., 2010 ▸; Coudray et al., 2011 ▸); the introduction of automated data-acquisition and processing software (Suloway et al., 2005 ▸; Stagg et al., 2006 ▸; Mastronarde, 2005 ▸; Li et al., 2015 ▸); the introduction of image-processing software suites with convenient graphical user interfaces (GUIs; Tang et al., 2007 ▸; Hohn et al., 2007 ▸; Baxter et al., 2007 ▸; de la Rosa-Trevín et al., 2013 ▸); and the development of integrated software environments and file formats that allow convenient interchanging between the different programs (Lander et al., 2009 ▸; de la Rosa-Trevín et al., 2016 ▸; Marabini et al., 2016 ▸). It is interesting to note that a similar development towards automation and ease of use happened in the related field of macromolecular structure determination by X-ray crystallography during the 1990s and 2000s (Blundell et al., 2002 ▸).
Instead of immediately executing a job once all its parameters have been selected, the user can also use the ‘Schedule’ button to schedule a job for execution at a later stage. Even though the files for the output nodes of scheduled jobs do not exist yet, they will still be added to the .Nodes/ directory that is used by the ‘Browse’ buttons on the parameter input fields. Thereby, once a job is scheduled one can select its expected output nodes as input for another job, which can then also be scheduled. In this manner, one can schedule multiple, consecutive jobs for future execution.
We describe how a pipelined approach to cryo-EM structure determination formalizes the concept of a directed acyclic graph in RELION-2.0. The new approach improves the user experience by reducing the scope for errors, automating bookkeeping, standardizing the organization of output files, allowing straightforward browsing through the project’s history and providing convenient disk-cleaning functionalities. The functionality to schedule multiple, consecutive jobs for future execution, and the option to execute these jobs in iterative loops, allows on-the-fly processing of the data while they are being acquired. This will lead to more efficient use of microscopy time, as more informed assessments of data quality can be made while the data are still being acquired. Finally, the possibility of exporting and importing (parts of) pipelines will allow the exchange of data-processing strategies between users, which will further improve the accessibility of the technique to inexperienced users and facilitate the development of standardized data-processing procedures. RELION-2.0 is open-source software. For download instructions and documentation, please refer to the RELION Wiki pages at http://www2.mrc-lmb.cam.ac.uk/relion. For user questions and feedback, please use the CCPEM mailing list at http://www.jiscmail.ac.uk/CCPEM.