Johan Turkenburg
| Venue: | Sheffield University |
| Dates: | 8-9 January 1999 |
| Programme Organisers: | Leo Brady and Johan Turkenburg |
Data collection and processing is obviously a crucial step in the determination of any X-ray crystal structure, as it provides the observations that are used throughout the process of structure solution and refinement. Mistakes made at this stage will haunt the unfortunate crystallographer, and may well block the road to crystallographic success. Computer programs can be run time and time again, but restraints on the availability of crystals and (synchrotron) data collection time mean that the data may only be collected once. In recognition of these facts of a crystallographer's life, 'Data collection and Processing' is a recurring theme for the CCP4 study weekends, and was chosen as the subject in 1987 and 1993. Since then third generation synchrotron sources have become available. The development of new optics systems for home sources, the more widespread availability of image plate detectors and more recently CCD detectors and the now universal application of cryo techniques, has changed the face of data collection. Together with advances in data processing and other crystallographic software, and the increased speed of computing this has led to a considerably higher throughput in macromolecular crystallography. It was therefore felt that it was high time to revisit this subject in 1999.
The meeting was held at the University of Sheffield. Thursday night a large number of delegates found their way to the bar in one of the Halls. A great deal of crystallography and other material was discussed while enjoying some beverages. Friday at 11:00 the meeting started in earnest. The programme promised a total of 18 speakers, covering all aspects of data collection and processing. Michael Rossmann gave a general introduction to data processing, and retold some of the history of the software used nowadays. He then went on to describe programmes more recently developed in Purdue in collaboration with ADSC. This has resulted in autoindexing routines which are extensively described in the literature and now also form part of the MOSFLM suite, as described in a later talk by Harry Powell.
Elspeth Garman talked about cryocrystallography and data collection in general. With her vast experience in this area she had examples of almost everything you should think about when preparing for an X-ray experiment, and of lots of more and less common mistakes. Surely no-one is going to breathe on their precious, frozen crystal ever again. She also discussed the potential of cooling crystals to liquid helium temperatures, and the benefits this might bring. This subject was returned to in the general discussion on Saturday afternoon.
After coffee Peter Lindley described the use of a third generation synchrotron source, and more specifically what is available at the ESRF. Colin Nave then explained how to match the X-ray source, optics and detectors to the protein crystal under consideration. Although there is often a limited choice in practice, it is obviously very important to be aware of the issues involved, especially when trying to decide which synchrotron to apply to for beam time. CCD detectors are now becoming standard equipment at most synchrotrons. They are, however, considerably more expensive than image plate detectors, often a major consideration for home laboratories. Steven Muchmore from Abbott Laboratories talked about his experience with CCD detectors on a laboratory source, and the pros and cons of such a setup. An exciting new development in the field of X-ray sources is the microfocus tube developed by Ulli Arndt and more recently Bede Scientific. Anne Bloomer has been testing such a source, and presented results as well as the potential for future development.
Over the last five or so years most laboratories have invested in mirror optics for their protein setups, often in the form of Yale mirrors. The next development is already upon us, with the introduction of multilayer optics for home sources. After tea Joe Ferrara from MSC presented results of extensive tests carried out with Yale mirrors and various incarnations of multilayer optics. The subject then changed from hardware to software, and Harry Powell described the implementation of the Rossmann indexing software in MOSFLM, as well as the algorithms involved. Andrew Leslie covered the next step in data processing, the integration of the diffraction data and described how the sigmas are determined for individual reflections. Zbyszek Otwinowski closed the day on a philosophical note, stressing that scaling, merging and statistics can be very case specific and do not follow hard and fast rules. Critical thinking throughout the process is of the essence.
Saturday morning most of the delegates returned after what some regard as the most important part of the CCP4 Study Weekend: the extended discussion time during dinner and in the bar. Zbyszek Dauter gave a very clear talk on data collection strategies. Reciprocal space suddenly seemed almost real. That diffraction is something we should consider in three dimensions was also stressed by Jim Pflugrath. Fine sliced data collection and matching data processing were illustrated with pizza slices and bits of pepperoni. Surely we all felt that this is a more tasteful approach to data collection! Martin Walsh talked about MAD data collection at the APS. The speed is certainly awesome, and in some cases structures can be determined in a matter of days.
After the coffee break Dominique Bourgeois illustrated the need for specialised processing software when processing time-resolved data. Algorithms developed here can also be used to alleviate overlap problems in conventional data sets. X-rays are of course not the only radiation which can be used to study structure. Johnny Grimes gave some examples of how EM data can complement the X-ray information, and how it can provide initial, low resolution phases. For many years twins have been a common phenomenon in small molecule studies. In recent years protein crystallographers have become more aware of their existence, and Naveen Chandra described various kinds of twinning as well as how to recognise them.
After another excellent lunch Randy Read showed how structure factor statistics may be used to detect outliers in non-redundant data. In such cases there are often at most two observations for a reflection, and deciding whether one of these is an outlier can usefully rely on information from other sources, like known distributions of structure factors. In the 'hot news' slot of the programme, Anastassis Perrakis described a beamline for microcrystals at the ESRF. The design of components like the goniometer and the retractable backstop should make dealing with such crystals more feasable. It is even envisaged that in an effort to overcome problems associated with the finite lifetime of frozen crystals in very intense beams, a number of crystals sitting in one cryo loop can be used one by one for data collection.
The meeting was closed by Phil Evans, who summed up the requirements for proper data collection and processing, and led the final discussion. Elspeth Garman introduced the issue of crystal lifetime, and the possible advantages of using liquid helium. No consensus was reached as to which processes are responsible for crystal damage. There was also some discussion on the need for the design of better backstops to reduce background and the usefulness of fine sliced data collection. Phil remarked that synchrotrons are not just a place to quickly collect data, but also to carry out proper data collection experiments.
The full proceedings will be published as a supplement to Acta Crystallographica section D, sometime towards the end of the year.
As one of the programme organisers I would like to thank the conference office and audio-visual department of Sheffield University, Pat Broadhurst (Daresbury) and the CCP4 staff, especially Alun Ashton and Sue Bailey.