The Symmetry Database (https://symmdb.iucr.org/) forms part of the online edition of International Tables for Crystallography and gives access to databases of crystallographic point and space groups. These online databases expand and complement the symmetry information provided in the print editions of International Tables for Crystallography Volume A, Space-Group Symmetry [1] or Volume A1, Symmetry Relations between Space Groups [2]. The information in the database can either be retrieved directly or generated ‘on-the fly’ using a range of programs. Help pages briefly explain the crystallographic data and the functionality of the programs. The data and programs that are currently available in the Symmetry Database are arranged into three sections:

(i) Space-group symmetry: The data in Volume A are extended to include the generators, general positions and Wyckoff positions of all 230 space groups, including the 530 settings for the monoclinic and orthorhombic space groups listed in Volume A. If data are not available for a particular setting directly, an arbitrary basis transformation can be specified and the data will be transformed to this new basis. The Wyckoff positions are specified by the Wyckoff letters, multiplicities, coordinate triplets and site-symmetry groups. Optionally, the symmetry operations of the site-symmetry groups of any point (within the unit cell or specified by its coordinates) can be calculated. Different types of notation are used for the symmetry operations: they are presented as coordinate triplets, in matrix form, using geometric symbols (indicating the type and order of the operations, and the location and orientation of the corresponding geometric elements, and screw or glide components if relevant) and as Seitz symbols. Information is also available for the Euclidean, chirality-preserving and affine normalizers of the space groups.

(ii) Symmetry relations between space groups: The maximal subgroup data given in Volume A1 are extended to subgroups of arbitrary index for all the space groups, and series of isomorphic subgroups are available for indices up to 50 for orthorhombic, tetragonal, trigonal and hexagonal space groups and for indices up to 27 and 125 for cubic space groups. Interactive contracted and complete graphs of chains of maximal subgroups, including basis transformations and origin shifts for each step, can also be generated. In addition, data for supergroups of arbitrary index of all the space groups are provided. In contrast to Volume A1, where only space-group types of supergroups are indicated, in the symmetry database each supergroup is listed individually and specified by the transformation matrix that relates the conventional bases of the group and the supergroup. The subgroup and supergroup data can be transformed to the basis of the group, left- and right-coset decomposition calculations can be carried out, and Wyckoff-position splittings can be obtained along with the relations between the coordinates of the positions within the group and subgroup.

(iii) 3D Crystallographic point groups: The data for the point groups, presented in an analogous way to the space-group data, include generators, and general and special Wyckoff positions. The data can be transformed to different settings, thus enhancing and extending the data tabulated in Volume A. Clear and instructive visualization of the symmetry elements of the crystallographic point groups and their stereographic projections, including interactive 3D polyhedra representations of idealized crystals, is also provided [3].

The Symmetry Database is available to all subscribers to the online version of International Tables for Crystallography. A Teaching Edition of the Symmetry Database, which can be used to obtain and explore the data for a selected set of space groups is also available online.

The Symmetry Database has been developed as part of an ongoing project between International Union of Crystallography, eFaber Soluciones Inteligentes SL. (Bilbao) and the Bilbao-Crystallographic-Server team. Most of the additional crystallographic data for the space groups, their subgroups and supergroups, and program algorithms have been provided by the Bilbao Crystallographic Server (www.cryst.ehu.es).

[1] International Tables for Crystallography (2016). Volume A, Space-Group Symmetry, 6th ed., edited by M. I. Aroyo. Chichester: Wiley.

[2] International Tables for Crystallography (2010). Volume A1, Symmetry Relations between Space Groups, 2nd ed., edited by H. Wondratschek & U. Müller. Chichester: John Wiley & Sons.

[3] Arribas, V., Casas, L., Estop, E. & Labrador, M. (2014). Comput. Geosci. 62, 53–61.

The ARP/wARP software provides automated model building in macromolecular crystallography and cryo-EM maps for structures of proteins, and their complexes with nucleic acids and small molecule ligands. The ARP/wARP remote service for macromolecular model building has been available since 2004 and was used to provide tens of thousands model building jobs remotely submitted by more than 4,000 users. A comprehensive description of the ARP/ARP web service, including a historical perspective will be provided. To allow the user a direct monitoring of the model building task, its progress and accumulated results are displayed graphically (e.g. the Wilson plot, the development of crystallographic R/Rfree-factors, the number of residues built) and in a tabular form as well as JavaScript-based cartoons of the built structures. The output files can also be downloaded when the job is completed. A user can rerun jobs with modified parameters and the results of these can be compared to each other. The analysis of the accumulated data and a number of take-home messages will be presented.

Macromolecular crystallography instruments around the world run more and more in a remote access or unattended configuration. This leads to less contact between humans and the hardware as well as less awareness of software and hardware states. Beamline failures that were in the past routinely reported by humans are now missed and lost in the noise of other issues. On another hand there is a need to have a chain of triggers from the beamline failure to the call out of a synchrotron staff that can assess and fix an issue. Finally, although most facilities have constant monitoring tools such has text messages or emails on catastrophic failures like loss of vacuum or cooling in the DCM, they tend to not monitor less important values due to the incapacity of a human being to deal with excessive amounts of information including false positives. Here we present a beamline monitoring software that intents to monitor EPICS PVs as well as other systems via HTTP restful interfaces, database connections or on disk file analysis and report in a configurable way to systems such as Slack, Email, Signal/WhatsApp or others. The use of a Slack bot allows update of configuration notifications as well as query some beamline states remotely before a support remote connection is required. Concepts as beamline mode as well as custom notifications for different staff members as well as a dependency chain of failures help reduce the number of notifications to a level which can be dealt with. The expectation is that this will be part of the on call / callout system, monitor the beamline live for upcoming possible problems as well as provide a log of the beamline states for the last day(s).

Machine learning can be justified only if input descriptors are crystal invariants independent of accidental choices. To use a household analogy, the average color of human clothes can be the easiest descriptor to extract from images but cannot be seriously considered for learning reliable information about people. Similarly, no properties of crystals can be reliably predicted from ambiguous parameters of a unit cell and a motif. Since crystal structures are determined in a rigid form, they should be considered equivalent modulo rigid motion or isometry, which preserves all interpoint distances. Then crystals can be justifiably distinguished only by isometry invariants that are independent of a unit cell and are preserved under any translations and rotations. Though Niggli’s reduced cell is unique, it is discontinuous under atomic perturbations, which are always present in real crystals. This continuity of invariants is important to quantify similarities between near identical crystals obtained by Crystal Structure Prediction as approximations to energy minima.

All machine learning approaches implicitly assume that a target property continuously depends on a given input, for example similar crystals should have close values of their lattice energy. We experimentally tested that the lattice energy is discontinuous with respect to the density, powder X-ray diffraction and packing similarity (root mean square deviation as computed by Mercury). For example, many crystals detected as similar by the above tools have very different energies. The AMD invariants are not only theoretically continuous under perturbations but also satisfy continuity for energy learning: we experimentally identified a distance threshold d and a constant c such that any distance between AMD invariants smaller than d guarantees an energy difference smaller than c times d.

Standard machine learning tools were trained on AMD invariants without chemical data for 10 min and predicted the lattice energy with a mean average error of less than 5KJ/mole on a CSP dataset of 5679 crystals containing about 250 atoms per unit cell.

Distances between AMD invariants are computed so fast that the pairs of all 229K organic molecular crystals from the Cambridge Structural Database were processed overnight on a modest desktop. The AMD invariants were recently extended to a complete isoset that uniquely and continuously represents any periodic crystal and allows an explicit reconstruction of a crystal.

For over 40 years, the Collaborative Computational Project Number 4 in Protein Crystallography (CCP4) has maintained, developed, and provided an integrated Suite [1] of world-class software that allows researchers to determine macromolecular structures by X-ray crystallography and other biophysical techniques.

Traditionally, the Suite is operated via CCP4i(2) graphical user interface, available for all major desktop platforms. More recent developments include interfaces that offer users the convenience of crystallographic computing on mobile devices and access to cloud-based resources. There are several good reasons for exploiting the distributed computing paradigm in crystallography.

First, cloud-based solutions have become particularly appealing given recent advances in automated structure solution methods. Such methods are demanding for both computing power and various databases, making them less convenient for offline setups.

Second, the cloud model of operations relieves researchers from the burden of maintaining software locally, providing 24/7 access to always ready, tested, and updated software setup.

Third, cloud computing streamlines data management and logistics. Collected data may be put in cloud-based projects directly from synchrotrons, bypassing offload to user devices. Cloud projects can be shared in real-time between a team of researchers working from various geographic locations. This aspect has been particularly helpful at the virtual CCP4 workshops during the pandemic.

CCP4 currently provides two interfaces for online work [2]. CCP4 Online, started from automatic Molecular Replacement service “BALBES” in 2008, is a web portal allowing users to run in the cloud the molecular replacement and experimental phasing pipelines in the CCP4 suite. In 2020, CCP4 released an advanced online platform, CCP4 Cloud, featuring a full desktop experience online. CCP4 Cloud includes an HTML5 interface for most crystallographic tasks and allows to develop and maintain structure solution projects completely online using common web browsers on any modern platform, including mobile devices.

We will discuss the latest developments, achieved results, and future directions. Providing a global computing infrastructure for protein crystallography is now a feasible task; are we ready to accept it in practice?

[1] M. D. Winn et al. Acta. Cryst. D67, 235-242 (2011)
[2] E. Krissinel, V. Uski, A. Lebedev, M. D. Winn, C. Ballard. Acta Cryst. D74: 143-151 (2018)

All biological processes rely on the formation of protein-ligand, protein-peptide and protein-protein complexes. Studying the affinity, kinetics and thermodynamics of binding between these pairs is critical for understanding basic cellular mechanisms. There are many different technologies designed for probing interactions between biomolecules, each based on measuring different signals (fluorescence, heat, thermophoresis, scattering and interference; among others). Evaluation of the data from the binding experiments and its fitting is an essential step towards the quantification of binding affinities. Here, we present user-friendly online tools to analyze biophysical data from steady-state fluorescence spectroscopy, microscale thermophoresis and differential scanning fluorimetry experiments. The modules from our data analysis platform (spc.embl-hamburg.de) contain classical thermodynamic models and clear user guidelines for the determination of equilibrium dissociation constants (K_ds) and thermal unfolding parameters such as melting temperatures (T_ms).