Minutes of the Working Group 2 Meeting -------------------------------------- Wednesday Jan 26th 2000, Birbeck College London ----------------------------------------------- Present: Phil Evans (Chairman) MRC, Cambridge Eleanor Dodson York Jan White Sheffield Alexei Vagin York Misha Isupov Exeter Ian Tickle Birkbeck Sheila Gover Oxford Paul Fyfe Bristol Paul Emsley Glasgow Jim Raftery Manchester Helen Saibil Birkbeck Robert Esnouef Oxford Garib Murshudov York Christine Cardin Reading Harry Powell Cambridge James Nicholson Daresbury Jim Naismith St Andrews Neil Isaacs Glasgow Martin Noble Oxford Sue Bailey Daresbury/CCP4 Alun Ashton Daresbury/CCP4 Peter Briggs Daresbury/CCP4 Martyn Winn Daresbury/CCP4 David Brown Daresbury/CCP4 Apologies: Quan Hao, Ravi Acharya, Lindsay Sawyer, Joachim Jaeger, Liz Potterton, Chris Gilmore 1) Minutes of last meeting and matters arising i. There was still no update on IMAGECIF from Phil. It was generally agreed that hardware & software manufacturers need to adopt this as standard. It was suggested that Harry could visit Bob Sweet (who has an IMAGECIF library). Another issue is that wavelength is not always written to image headers anyway when collecting data at beamlines. ii. Paul Emsley was asked what has happened with Chris Gilmore's maximum entropy program EM. Paul thought that no progress had been made since Wai finished, but he would contact Chris to find out more. iii. Phil thanked the organisers of this year's study weekend. 2) Update of status of EM programs Helen Saibil reported that following consultation exercises there is a consensus within the EM community that harmonisation of EM software (using CCP4 as a model) is highly desirable. Authors of various software packages have expressed an interest in being involved once the harmonisation process has been initiated. The first issue is standardisation of image formats. There are a number of different packages for different aspects of EM, many using the MRC image format (similar to CCP4 map format), and some not (e.g. IMAGIC). There is a website which explains the problems with different formats (http://www.EMBL-Heidelberg.DE/ExternalInfo/fuller/ccp4.html). The second issue is deposition of EM data. Currently EM images are deposited with the BIOIMAGE database but this will soon be withdrawn, so the community wants some way of depositing EM data with the PDB. It was suggested that turning the density maps into structure factors would be useful both for storage and for possible deposition. The progress with harmonisation made so far: Steve Fuller has completely converted his programs to use CP4 formats. The Plot82 libraries have been updated to output postscript. Richard Henderson & Tony Crowther have agreed to update their libraries to be compatible with CCP4 (not done yet). For the future: Steve Fuller wants to deposit his programs with CCP4 for distribution, but he would continue to support and maintain it himself. It was agreed that this could serve as a model for the general management of EM software through DL. User problems would need to be passed onto the program authors e.g. via some equivalent to the current ccp4@dl e-mail address. An EM equivalent of ccp4bb already exists (3dem) for general queries. There will also need to be minor changes to CCP4 maplib, (e.g. to deal with EM maps being either 3-d images or a stack of 2-d images) and SFALL (which seems to have a limit on the size of temperature factors). Phil requested a list of explicit changes that need to made to CCP4 software from the EM people. 3) Study weekend 2001 Working Group 1 had suggested 'Molecular Replacement and its Relatives' as the title with Jim Naismith and Kevin Cowtan as the organisers of the 2001 study weekend. The meeting accepted the suggestion but noted Kevin's absence from York from the end of September to late December. It was commented that part of the Friday programme could be devoted to discussion on the EM and X-ray topics raised in the last study weekend and also that the content of some talks should be toned down to student level. The latter found acceptance with the meeting and it was agreed that careful selection of speakers was required and should be brought together at the May meeting of this group. Suggestions on topics and speakers should be passed to the organisers. 4) Progress with NT Port Alun reported that the low-level libraries were now ported OK except for the binsort routines; bypassing these meant that all but six of the programs compiled and ran. It was suggested that NT must have some kind of system binary sort, cf VMS. Dave Love had been asked for help but was unable to, due to other commitments. Jim Naismith suggested asking Paul Adams, since CNS also runs under NT and so they may already have solved this problem. Liz still needs to make changes to the gui to accommodate NT but this taking longer than expected; in the meantime Alun announced there is a DOS/Cynus version of the suite which runs with scripts, available for testing by willing users. 5) Technical support for user groups Alun mooted the idea of a half-day session attached to a workshop, teaching people how to set up a computer to use crystallographic software. This would address a need particularly in small academic groups where people are less able to rely on centralised computing facilities, and so have to take responsibility for their own computing support. Counter to this it was argued that a single workshop would be of limited short-term benefit, and that information might be more usefully presented via the web. Alun pointed out the benefits of the "hands-on" nature of a workshop, and noted that a webpage would require a high level of maintainance to prevent it dating quickly. Another suggestion was to use the webpages as a repository for advice posted via the bulletin board. (Maria Turkenburg is already extracting summaries generally from messages). The O-bulletin-board was suggested for questions about hardware. 6) Progress with REFMAC Garib is working on a new version which will incorporate Martyn's TLS and Alexei's dictionary, which will make PROTIN obsolete. The dictionary can take coordinates and generate connectivity lists, but it still needs testing, since in certain cases (e.g. absence of hydrogens) it makes mistakes. After these changes the next major change is to introduce bulk solvent correction - this has been used successfully in York already, and a beta-release should be available next month (Feb). After that Garib wants to concentrate on new projects (e.g. anomalous refinement). 7) Details of release 4.0 Martyn reported that CCP4 version 4.0 is now available. New features include the CIF libraries, and five programs have harvesting capabilities. "Entry" programs assign default project- and dataset names automatically, if these are not specified by the user. Phil commented that SFCHECK should also have harvesting capabilities in future. In principle the EBI is ready to accept the harvest files, but this procedure is not yet fully automated, and so it is recommended that individual depositors contact the EBI when beginning deposition. 8) Development of 3D graphical viewer WG-1 has decided that CCP4 should develop some form of 3-d graphical viewer for map and model display, which could run under the current gui. The motivation for such a project was to fill a gap in the functionality offered by the suite; also from the need to keep CCP4 in continual development, and for CCP4 to be more proactive in meeting the requirements of the user community. Such a viewer could have a number of forms ranging from: display only; map fitting/interpretation; a general display & rendering program. The questions were: what features should be specified? and how such the development of such a project be undertaken? Many issues were raised and discussed. Generally people seemed to be thinking of replacements for currently-available model-building packages e.g. O (criticised as out-of-date) and QUANTA (too expensive). Two approaches could be considered - writing a new package from scratch, or developing an existing package in a direction determined by CCP4. Phil pointed out that in the latter case, groups would be unlikely to give away years of research by selling their existing package effectively for a year's pay. The issue of licensing arose. This was considered to be a top priority. The software would need to be licensed to CCP4 in such a way that it could be freely distributed. Many people expressed the opinion that they rather have the job done as a research project, instead of it being tendered to a commercial company. WG-1 hasn't placed restrictions on how much could be spent on the software, and Jim Naismith pointed out that some of CCP4's industrial customers were prepared to provide extra funds (in addition to the annual subscription) for development of such a package by CCP4. However, it was suggested that even then the cost of commercial development might be a lot higher than CCP4 would be able to afford. It was also pointed out that the package would also require support, not just developement. Martin Noble and Robert Esnouef were asked to comment on the technical details of developing a package from scratch. They agreed that the most important thing initially was decide the ambition for the package, after which it should be relatively quick (eg 18 months) for someone to design an appropriate data structure, develop the core program and libraries for displaying maps etc. which could act as a usable frame for other developers to add their own modules. There was also some discussion about how this might impact on the rest of the suite; for example, would this be a good point to revise/rewrite the suite under an object-oriented framework? The "modular" approach might offer a way to migrate from Fortran to (e.g.) C/C++. Neil Isaacs suggested that advice from specialists was needed, and that a meeting of interested partners was the best first step. He warned people not to concentrate on the mechanics of funding, at this stage the most important thing is to arrive at a technical specification for the software. It was decided that Phil should organise a meeting for interested parties (no more than 20 people, mainly potential developers) before the next WG2 meeting. Sue should survey CCP4 users via the BB a few weeks before the meeting, to ascertain which packages are in popular use, good/bad features and possible enhancements. It was also suggested that industrial users be polled separately for their opinions. ACTION: Phil, Sue 9) GUI update A new version 1.1 of the interface was released with v4.0 of CCP4, with a number of new tasks. It can now also handle harvesting. The current major omissions are: handling of NCS, analysis of protein structure, and pdbset interface (which should be in the next version). There are still problems with porting the GUI to NT (handling of filenames, using sockets rather than the "send" command, and some inefficiency when starting a job). All these will need significant changes to put right, after which volunteer testers will be required. 10) MODLIB and rationalisation of maths routines Peter Briggs outlined plans to rationalise the mathematical subroutines in the suite (both in programs and in libraries). Routines currently found in the MODLIB library have now been documented. A survey of the programs revealed much duplicated code and functionality, and highlighted gaps in the library. The plan is for duplicated code to be removed and useful routines to be added to the library, and in future to incorporate freely-available BLAS/LAPACK libraries into the suite, for improved efficiency and robustness. 11) Progress with MOSFLM Harry reported that postrefinement with fine phi-slicing is working ok in MOSFLM provided that there are plenty of spots, but needs improvement for harder cases (this wil happen over the next few months). There is a new jiffy for estimating mosaicity of image, though this also has some problems. There has been some debugging of the autoindexing, and backstop fitting has been improved. Harvesting is now incorporated so the CCP4 v4.0 libraries are needed. A beta version is available now, and the next full version is due out by Easter. 12) Date of next meeting May 3rd 2000 at York University.