New versions of iMosflm (1.0.3) and Mosflm (7.0.5) are now available from the MRC LMB ftp & web sites (10th August 2009). Changes since 1.0.0 / 7.0.4 Enhancements to the iMosflm GUI (Version 1.0.3) =============================================== New features and improvements: * Direct beam search during indexing. * Pointless can be run multiple times as images are being integrated * Accuracy of Strategy prediction improved. * Detector parameters can be reset with one button. * Improved mosaic spread refinement during Cell Refinement. * "Find hkl" and "Pixel values" options in image display * Many bug fixes to improve stability of imosflm. Details: 1. Direct beam search: If the direct beam position is uncertain, imosflm can do a grid search, and will display the indexing results for the triclinic solution for each beam position tried. The rms error in spot positions from the indexing can be used to select the best solution. Double clicking on a solution will repeat that indexing, giving the usual list of possible solutions. The direct beam search is started by clicking on the text "Start Beam-centre search" in the Autoindexing pane and, once started, aborted by clicking on "Abort Beam-centre search". The results can be collapsed or expanded using the [-] and [+] symbols. By default, +/- two steps of 0.5mm about the initial beam position are searched, but these parameters can be changed by selecting the "Indexing" tab in "View/Processing options". For large unit cells the results will be ambiguous, but can be distinguished on the basis of integrating a few images and running Pointless. 2. If the initial space group assignment at the Autoindexing step is uncertain, Pointless can automatically be run multiple times as the images are being integrated to check the spacegroup. Select the "Processing" tab in "View/Processing options" and click the tick boxes for the options "Multiple MTZ files (1 per block)" and "Feed each MTZ file to Pointless". With the first option, the results of integrating each block of images (typically 8-12, can be set manually in the same tab) is written to a separate MTZ file, derived from the MTZ filename displayed in the Integration pane by adding the suffix _001, _002, _003 etc to successive blocks (eg myxtal_001_001.mtz, myxtal_001_002.mtz ...). With the second option, once completed these files are automatically passed to Pointless and the results displayed in a Web browser. The Pointless page in the browser can be manually updated (ie refreshed ... this does NOT happen automatically) to view the results with an increasing number of images included. The summary gives the number of batches (ie images) used. 3. The size of the unit cell is reduced in order to speed up the strategy calculations. This could result in some inaccuracy, for example the anomalous completeness could be shown as higher than overall completeness. The shrinkage factor has been reduced to improve accuracy, and on modern machines the calculation is still very rapid. 4. If an attempt is made to integrate or refine the cell with an incorrect indexing solution, the detector parameters (tilt,twist,yscale,distance,beam coordinates) can refine to inappropriate values. The default/original settings can be restored by selecting the "Detector" tab in "View/Experiment settings" and clicking on the "Reset Default Parameters" button. 5. If the initial cell parameters from Autoindexing are inaccurate, subsequent Cell Refinement (or integration) can lead to the mosaicity refining to a very small (or negative) value, especially if the images are high resolution (beyond 2.5A). To help prevent this happening, during the first pass of a Cell Refinement run the input mosaicity is not refined until all images have been integrated and the cell parameters are refined. In subsequent passes (the images are re-integrated if there is a large shift in cell parameters), the mosaicity will be refined for each image as before. If the mosaicity still refines to an inappropriately small value, it can help to reduce the resolution to ~2.5A initially, then repeat the refinement at higher resolution. Further investigation is currently in progress. 6. "Find hkl": To locate a particular reflection, select "Find HKL" from the "Tools" menu bar in the Images display. Enter the indices and click on "Ok". Note that at present other window options are frozen while the Find hkl window is open, this will change in the next release. 7. To examine pixel values, click on the "Selection" icon (black arrow) and press middle mouse (right mouse button on Macs). Pixel values in a box of 11x11 pixels is displayed in a new window (this size cannot be changed yet). Enhancements to ipmosflm ======================== 1. Treatment of spots where part of the measurement box lies in an inactive part of the detector improved. 2. For Pilatus images, correctly exclude spots lying in the gaps between tiles from the strategy calculations done in imosflm. 3. Improved treatment of images from NOIR detectors, minor issues with Rigaku CCD images dealt with. 4. Improved stability (many bug fixes) and speed. ---------------------------------------------------------------------- Changes since 7.0.3 Enhancements to the iMosflm GUI (Version 1.0.0) =============================================== A greatly improved and extended version of the iMosflm GUI. This now has almost all the functionality of the old X11 based ipmosflm GUI and is much easier to use. We strongly recommend users switch to the new iMosflm GUI as the X11 version is now inferior and will no longer be supported. For expert users, iMosflm now has the option of specifying ipmosflm keywords for the very specialised options that are only rarely used and are not yet included in the iMosflm GUI explicitly. A "wait" option allows images that have not yet been collected when the iMosflm session is started to be included in an integration run (previously this was not possible). This is useful when processing data as it is collected at a synchrotron (or on a home source). The "testgen" option, which calculates the maximum oscillation angle possible while avoiding spot overlaps, or reports the number of overlaps for provided oscillation angles, has been fully implemented. It is possible to run both POINTLESS and SCALA directly from the iMosflm GUI. Results are reported in a web browser using Baubles. In both cases, the programs can only be run in default mode (no user input is possible). A new iMosflm User Guide is available and the iMosflm Tutorial has been updated. Enhancements to ipmosflm ======================== Significant improvements in the spot finding routine when dealing with very weak images or very long unit cells resulting in poorly resolved spots. The improved spot finding increases the success rate of the autoindexing. and some changes to the indexing itself has further improved the reliability of indexing. New detectors such as the Pilatus, Saturn, Jupiter, Oxford Diffraction and Bruker CCDs can all be handled better than previously. Many bugs fixed Changes since 7.0.2 This is a bug-fix release which addresses some minor (but annoying) issues which have come to light since the release of version 7.0.2. The following bugs have been fixed in Mosflm itself. Other issues that arose in iMosflm have been addressed separately: * Image display for R-Axis detectors was reversed in iMosflm (the new interface). * The presence of extremely large pixel values (>262128) for Mar IP scanners caused Mosflm to shut down. * Direct beam co-ordinates read from Mar CCD image headers not sited at ESRF had X and Y swapped. * Direct beam co-ordinates were not used from Ed Westbrook NOIR images by iMosflm (although they were used by Mosflm). * Reversephi check box in iMosflm did not function as intended. Changes since 7.0.1 The main purpose of this new version is to provide enhanced support for the new interface, iMosflm (currently version 0.6.0). * New detectors - Mar555 Flat Panel. Pilatus "miniCBF" images added. * Better support for Bruker 100 Format images (converted with frm2frm). * Change IP address for communication with iMosflm to the loopback address (127.0.0.1 or "localhost"). * Rectangular non-square detectors added for output to iMosflm display. * Bug involving R-Axis detectors writing d*Trek style headers fixed. * Many small bugs fixed. Changes since 7.0.0 * New detectors - "native format" Pilatus 6M (i.e. not "miniCBF) and ADSC Q270 * Better support for CBF files produced by ADSC detectors. * Indexing made more robust for difficult cases. New keywords added for command-line indexing. * Native MS-Windows version now available (not relying on Cygwin environment). * Native Mac Intel build. Statically linked so it is not dependent on a specific version of libgfortran.*.dylib being installed on the user's machine. * Check added to make sure the current image has finished being written. * Many small bugs fixed. Changes since 6.2.6 * Many changes which allow processing information to be communicated between Mosflm and the new TclTk-based Graphical User Interface, which is now called iMosflm. Mosflm itself can be run as before as a standalone program, but can also be run from iMosflm. The original X11 GUI, based on John Campbell's XDL_VIEW libraries is still included and can be used if desired. * Now checks to see if an image file is still being written to before attempting to process it. * More improvements to the autoindexing to make it more robust in cases where it failed previously. * Build developed which will run (without the X11 graphics) on MS-Windows, allowing iMosflm to run properly on that operating system. It works on both XP and Vista, and may well work on 2000 and 98. * Many small bug fixes which address rare but annoying problems. Changes since 6.2.5 Most changes are "behind the scenes" for improved performance, so there are no new major commands. * Improvements to spot finding for autoindexing, including a new mode "FINDSPOTS LOCAL", which uses a locally (rather than radially) calculated background. * Many changes for use with the new Graphical User Interface. * numerous small bug fixes and minor changes. Changes since 6.2.4 * Detector contribution to uncertainty in sigma(I) has been improved. * More types of excluded area can now be specified, either via the command-line (see major command EXCLUDE) or using mouse clicks on an image in the GUI. The new areas are o generalized convex quadrilateral o inside an arc region o outside an arc region o Ice or other powder rings can also be picked with mouse clicks. This has been accompanied by a tidying up of the "Beam/backstop" button in the GUI. * Autoindexing has been further improved, and is also more flexible and gives (optionally) more feedback. All options are available via either the GUI or comand line indexing. These include: o choice of minimum cell edge o L.S. refinement of basis triclinic solution prior to symmetry tests o Distortion index in addition to Penalty for each solution * The user can now choose either hexagonal (H3, H32) or rhombohedral (R3, R32) axes for rhombohedral cells. * New detectors added include MarMosaic 325mm and Ed Westbrook's lens coupled CCD detector (NOIR at APS). * minor improvements to STRATEGY option * numerous small bug fixes and minor changes. Changes since 6.2.3 * CCP4 version 4 no longer supported. Mosflm version 6.2.4 requires the prior installation of CCP4 version 5. * Revamped build, which should be easier to customize for different compilers. * Beam centres now read from MarCCD; improved processing of Oxford Diffraction and Bruker images. * Output option for Sasha Popov's intelligent strategy program BEST (available from him at sasha@embl-hamburg.de) - invoked by keywords BEST ON. * The bug which occasionally produced the message "QOPEN (no streams left)" has been fixed. * Many other small bug fixes and minor changes.