Research Article: Scaling diffraction data in the DIALS software package: algorithms and new approaches for multi-crystal scaling

Date Published: April 01, 2020

Publisher: International Union of Crystallography

Author(s): James Beilsten-Edmands, Graeme Winter, Richard Gildea, James Parkhurst, David Waterman, Gwyndaf Evans.


A new scaling program is presented with new features to support multi-sweep workflows and analysis within the DIALS software package.

Partial Text

The first major step in processing crystallographic data is the integration of diffraction images, i.e. the extraction of a set of intensities and error estimates from the raw data. These integrated intensities are a product of F2, the squares of the structure-factor amplitudes, and experimental effects: the aim of scaling is to correct for the experimental contributions to the measured intensities to give a set of intensities proportional to F2. Scaling will typically account for systematic effects such as changes in sample-illumination volume, beam intensity, secondary beam absorption and global radiation damage during a single-sweep measurement, which are multiplicative in their effect on the measured intensities. The scaling process also places the intensities from multiple sweeps on a common scale. Once scales have been applied to all reflections, symmetry-equivalent reflections can be merged together. Several programs exist to scale macromolecular diffraction data, including SCALA (Evans, 2006 ▸), AIMLESS (Evans & Murshudov, 2013 ▸), SADABS (Sheldrick, 1996 ▸), SCALEPACK (Otwinowski & Minor, 1997 ▸) and XDS/XSCALE (Kabsch, 2010b ▸).

This section describes the implementation details of the key components of the scaling algorithms and potential scaling workflows.

To validate the new scaling algorithms within DIALS and to demonstrate the use of the tools, the results of scaling a variety of experimental data sets are shown in this section, including a weak macromolecular data set, a multi-crystal narrow-sweep data set with significant radiation damage and a multi-crystal data set. The scaling algorithms in DIALS can be run using the command-line program dials.scale, with optional keyword parameters. The full commands used to analyse each example data set are given in Appendix B, and the majority of plots shown can be found in the HTML report generated by dials.scale.

Scaling algorithms have been implemented in dials.scale, which is part of the DIALS software package, incorporating well established scaling approaches and introducing specialized tools for multi-crystal data scaling, validation of scaling parameterization and optimization approaches, and a high degree of flexibility to its workflow. The effectiveness of the implementation has been demonstrated on example macromolecular crystallographic data.




Leave a Reply

Your email address will not be published.