A new software suite, called Crystallography & NMR System (CNS), has been developed for macromolecular structure determination by X-ray crystallography or solution nuclear magnetic resonance (NMR) spectroscopy. In contrast to existing structure-determination programs the architecture of CNS is highly flexible, allowing for extension to other structure-determination methods, such as electron microscopy and solid-state NMR spectroscopy. CNS has a hierarchical structure: a high-level hypertext markup language (HTML) user interface, task-oriented user input files, module files, a symbolic structure-determination language (CNS language), and low-level source code. Each layer is accessible to the user. The novice user may just use the HTML interface, while the more advanced user may use any of the other layers. The source code will be distributed, thus source-code modification is possible.The CNS language is sufficiently powerful and flexible that many new algorithms can be easily implemented in the CNS language without changes to the source code. The CNS language allows the user to perform operations on data structures, such as structure factors, electron-density maps, and atomic properties. The power of the CNS language has been demonstrated by the implementation of a comprehensive set of crystallographic procedures for phasing, density modification and refinement. User-friendly task-oriented input files are available for nearly all aspects of macromolecular (i') 1998 International Union of Crystallography Printed in Great Britain -all rights reserved structure determination by X-ray crystallography and solution NMR.

The determination of macromolecular structure by crystallography involves fitting atomic models to the observed diffraction data. The traditional measure of the quality of this fit, and presumably the accuracy of the model, is the R value. Despite stereochemical restraints, it is possible to overfit or 'misfit' the diffraction data: an incorrect model can be refined to fairly good R values as several recent examples have shown. Here I propose a reliable and unbiased indicator of the accuracy of such models. By analogy with the cross-validation method of testing statistical models I define a statistical quantity (R(free) (T) that measures the agreement between observed and computed structure factor amplitudes for a 'test' set of reflections that is omitted in the modelling and refinement process. As examples show, there is a high correlation between R(free) (T) and the accuracy of the atomic model phases. This is useful because experimental phase information is usually inaccurate, incomplete or unavailable. I expect that R(free) (T) will provide a measure of the information content of recently proposed models of thermal motion and disorder, time-averaging and bulk solvent.

Molecular dynamics was used to refine macromolecular structures by incorporating the difference between the observed crystallographic structure factor amplitude and that calculated from an assumed atomic model into the total energy of the system. The method has a radius of convergence that is larger than that of conventional restrained least-squares refinement. Test cases showed that the need for manual corrections during refinement of macromolecular crystal structures is reduced. In crambin, the dynamics calculation moved residues that were misplaced by more than 3 angstroms into the correct positions without human intervention.