Minutes of the Working Group 2 Meeting -------------------------------------- Wednesday May 3rd 2000, York University --------------------------------------- Present: Phil Evans (Chairman) MRC, Cambridge Eleanor Dodson York Jan White Sheffield Alexei Vagin York Sheila Gover Oxford Paul Emsley Glasgow Jim Raftery Manchester Garib Murshudov York Christine Cardin Reading Peter Briggs DL/CCP4 Quan Hao DeMontford Tadeusz Skavzynski Glaxo Wellcome Jim Naismith St Andrews Martyn Winn DL/CCP4 Alun Ashton DL/CCP4 Liz Potterton York Kevin Cowtan York Sue Bailey DL/CCP4 Maria Turkenburg York David Brown DL/CCP4 Charles Ballard DL/CCP4 Apologies: Chris Gilmore, Lindsay Sawyer, Paul Fyfe, Ian Tickle, Ravi Acharya, Peter Moody, Misha Isupov and James Nicholson 1) Minutes of last meeting and matters arising i. The new CCP4 programmer Charles Ballard was welcomed. ii. IMAGECIF: Harry Powell and Dave Love have discussed implementing this (although Dave suggets a different format, Nexus, as an alternative). The most important thing is that some standard (whether IMAGECIF or other) should be adopted as soon as possible, and pressure should be put on synchrotron staff to implement this. It was noted that at Grenoble staff are starting to automate the capture of wavelengths etc. If MOSFLM is updated to write out IMAGECIF headers then this might put pressure on DENZO to follow suit. A separate issue was raised by Jim Naismith regarding DENZO being the only way to prepare input of unmerged data for SOLVE; Phil agreed to contact Tom Terwilliger and ask about updating SOLVE to accept input from MOSFLM too. b. Chris Gilmore's Maximum entropy program "EM": Paul Emsley reported that at present the program doesn't work, but Chris has a new postdoc who is now trying to finish it. c. EM programs: Phil reported that Tony Crowther and Richard Henderson are still updating the libraries to be compatible with CCP4. Progress has been slow, but Phil felt that any initiative needed to come from the EM community and not from CCP4. 2) Study Weekend 2001 A provisional programme had been circulated by the organisers and found general acceptance by the Group. Six invitations had been issued to keynote speakers and one decline and three acceptances had materialised. Discussion took place on the merits of inviting some alternative speakers and bringing in new faces. A detailed review was made of each proposed programme session and minor revisions in sessions 3, 4 and 6 were taken on board by the organisers. It was accepted that Session 6 would be better placed before Session 4 at the start of Saturday. 3) Progress with NT port Alun reported that Garib has provided him with a binsort replacement written in standard Fortran, and although this hasn't been tested it should enable the porting of the remaining programs. Testing of the NT version under the GUI had uncovered some more problems which have now been fixed. He has also tested under Windows 2000 and SG NT. There will be a pre-beta release to selected sites before the end of May, with wider beta distribution hopefully soon after. The package will be a self-extracting/self-installing package. A separate discussion arose concerning whether Garib's new binsort should replace the existing binsort within the suite. It should be more portable and Garib believed it could also be made to operate faster than the current version. There was also a complaint that the current default memory allocation for binsort is very small, and Martyn agreed that this should be increased in future releases. 4) Report from meeting on CCP4 coordinate format 19-4-00 Since the original decision had been made (2+ years ago) to replace CCP4's pseudo-PDB working format with a subset of mmCIF, the EBI people have had more experience with the new format and aware of some of its problems/limitations, while CCP4 has to accommodate ideas for a data model for the proposed 3D viewer. Martyn reported that no firm decisions had been reached at the coordinate meeting, with most of the discussion being about the possible data model. He outlined the range of roles that a coordinate format needed to cover (e.g. refined coordinates, heavy atom positions, model fragments etc). The mmCIF dictionary could provide a basis for the new format though it would need the addition of new extensions (eg for harvesting), but the mmCIF relationships do not provide a good data model for CCP4 applications since it doesn't support a simple hierarchy. The proposal was to ignore the official mmCIF relationships and use our own. Possibilities were also suggested for the actual working file format. Binary files have a number of advantages from a programming and retreival viewpoint but were unpopular because people like to edit their files manually for various reasons. Tools could be created for manipulating the binary format in the same way as for ascii formats at present, though it might be difficult to provide the full range of functionality that users currently have access to. An alternative would be to dump the file contents to an ascii format for editing before regenerating the binary file. This presents other problems if the intermediate file is "lossy" (eg CIF to PDB) - this issue needs to be addressed anyway for CCP4 programs to be able to communicate with external programs such as O. Proper import and export facilities would be required. Alexei already has routines in MAKECIF which can read and write information transparently from different formats. It was also suggested that CCP4 should talk to other software developers about these issues before proceeding. The main perceived disadvantage to using an ascii format are the possible overheads associated with reconstructing relationships between data items when reading them into a program. It was not clear whether this would be prohibitive, and it was agreed that further investigation was required before further decisions are made. The next steps are: to decide on the data model and a set of data objects/definitions (which Kim is already doing). This can be started independently of the 3D viewer but interaction will be required later on. The ideas will be implemented by the new EBI/CCP4 position, when appointed. 5. Development of 3D graphical viewer Alun distributed preliminary results from the user questionnaire posted recently to the bulletin board, and gave some details about the meeting planned for the 15th May at York. The meeting will be invitation only, to reduce numbers (currently 15). A number of people developing packages were approached but some (eg Alwyn) had declined the invitation, seeing this initiative as direct competition. Some of issues were clarified: it was suggested that there was no point in simply developing a 3D viewer when it is already easy to write scripts to lauch existing software eg from ccp4i. The new package should be something like Quanta, but with the source code freely and publically available, with no restrictions on distribution. This would encourage a "community effort" in developing and extending the new application which should mean that it grows quickly. At least one other group (Thomas Hamelryck & Morten Kjeldgaard) seem to be working on a similar project to the CCP4 initiative (which would also ultimately be freely available) but seemed reluctant even to attend the meeting. It was suggested that we should approach them again, since we would prefer to work with others if their interests coincide with ours. However Alun pointed out that there are already a lot of other people who are interested in working with us - so it would be realistic for CCP4 to go it alone. There was some digression over copyright and licensing issues. The meeting generally favoured making the viewer GPL (General Public Licence i.e. freeware). There are problems with doing this with existing CCP4 programs because CCP4 does not have full claim to the copyrights on many of them. Another potential problem (for the viewer project) is that intellectual property rights would normally belong to the institution employing the program author(s), and not to the authors themselves. A separate issue was the recent legislation in some American states which voids "no warranty" clauses such as that used by CCP4, and could potentially leave CCP4 open to legal action. It was suggested that these problems could be avoided by having "configure" ask the user if they agreed to the no warranty clause before proceeding with the installation of the suite. It wasn't clear if this would completely resolve the issue however. 6. Progress with REFMAC Garib reported that a pre-beta version of REFMAC (refmac_5.0.*) is now available from the York ftp site. There are a number of new features: (i) bulk solvent modelling plus Babinet correction - it is now possible to have both the Babinet correction _and_ the explicit bulk solvent modelling (Roberto Steiner) (ii) TLS (Martyn Winn) (iii) new dictionary (Alexei Vagin). Regarding bulk solvent modelling, Garib reported that the first stage is finished but suggested that CCP4 could fund Roberto to continue working it. He also talked about the new dictionary, which is very flexible and recognises a number of different formats (eg. PDB, CIF). New refmac documentation is also coming soon, this will include detailed information on input, output and log files plus links to sites maintained by Garib for the background theory and for new dictionary entries. It was suggested that users might be able to deposit their own dictionary entries, provided that some form of validation could be used. Kevin suggested that as refmac is currently CCP4's flagship program, the new version should be advertised as widely as possible once it is properly released, and that there should perhaps also be a separate refmac website. 7. GUI Update Liz gave some details of her molecular sketcher program which should interface with the libcheck program (itself an interface to the new dictionary), for creating new dictionary entries. It can read in and display structures (PDB, CIF etc) or existing dictionary entries, and allows the user to modify atoms, links etc. She would welcome any suggestions about other features(?) that are required for interactions with other programs. Liz has also been talking to Harry and Andrew regarding updating the MOSFLM interface with Tcl/Tk and tools from ccp4i. She has demonstrated that Tcl/Tk graphic capabilities could be used for these purposes but the main problem will be separating the core MOSFLM code from its current user interface. 8. Date of next meeting (Provisionally) Wednesday 20th September 2000 at Birkbeck College London. 9. AOB i. Paul Emsley announced that version 0.4.4 of Chart is now available. ii. Sue announced that the EU grant proposal to fund workshop organisation and support had been successful. This would be used to help fund workshops like Como plus "integration meetings" (eg integrating software at synchrotrons), and if people have ideas for workshops which require funding then they should contact her.