################################################################### # # This is the latest (March 02 2016) snapshot release of the BUSTER # package (BUSTER, rhofit, gelly, grade, buster-report and # pipedream) for academic users. # ################################################################### # ################################################################### # # Copyright (C) 2003-2016 by Global Phasing Limited # # All rights reserved. # # This software is proprietary to and embodies the # confidential technology of Global Phasing Limited # (GPhL). Possession, use, duplication or dissemination of # the software is authorised only pursuant to a valid # written licence from GPhL. # # Authors: G. Bricogne, E. Blanc, M. Brandl, C. Flensburg, P. Keller, # W. Paciorek, P. Roversi, A. Sharff, O. Smart, C. Vonrhein, # T. Womack # ################################################################### =================================================================== Changes since the last BUSTER stable release Mar 02 2016 =================================================================== BUSTER improvements and fixes (released Mar 02 2016): * Instead of only warning about a model being too far away from the origin (often resulting in a failure later), the pdbchk step will now also correct for this. By default the required shift will be undone at the end of refinement so that the output PDB file will have the same position as the input model. * During refinement, the same set of automatically determined EXCLUDE cards is now used throughout the refinement. These are determined during big-cycle 1 to avoid bad contact terms between atoms across symmetry elements. By re-using the same set throughout we avoid that a shift in coordinates - to accommodate electron density features - could result in an initially excluded contact term to suddenly becoming active (which could result in a sudden movement at a later big-cycle). Pipedream improvements (released Feb 25 2016): * By default, the post-refinement step will refine the occupancy of ligands fitted by rhofit. * Added -nooccref command line option to tell post-refinement not to refine ligand occupancy * Added -postthorough option. Tells pipedream to perform a more thorough post-refinement, involving two BUSTER runs. * Added -nobr option to instruct Pipedream to not run buster-report * Updated Phaser version warning message for new version of Phaser in CCP4 7.0 * Improvements to ERROR reporting BUSTER improvements and fixes (released Feb 25 2016): * pdbchk_element: update to handle components.cif as of 20160112. * fetch_PDB: take care of structure-factor CIF files where the _diffrn_radiation_wavelength loop might also contain a weight (.wt) item: see eg 5ewn. BUSTER improvements and fixes (released Dec 14 2015): * Major reworking of geometry checks, including the initial sanity check: * When encountering atoms without bonds, some additional analysis is done to help users when this occurs for known/standard residues. * Re-organisation of the initial sanity check routine - keeping the underlying logic, but * tidied up formatting of messages; * adding an identifier (check-NNN) to each check - visible only in the detailed gelly_sanity_test.log file and not in the shortened gelly_sanity_test.lis version. This helps communication with developers but also detecting common patterns or problems. * sanity check is now aware of Buffer, RNA/DNA, Metal and Sugar residues; this allows much finer control of what message to produce; The biggest impact on reducing the number of close contacts often wrongly classified as errors is due to a second check once such a close contact is detected. Instead of assuming anything closer than an ideal contact is most likely a covalent bond we now analyse the distance of those two atoms (knowing their atomic type) against ranges of known covalent distances for those two atom types. If the distance is within -0.1/+0.1 A of that range we will still give an ERROR, but if it is only within -0.5/+0.5 we now give a WARNING. Anything further away is only reported. * Added support to compute and analyse anomalous Fourier maps (if anomalous differences are present in input MTZ file). This can be switched on with BusterAnoFourier=yes (default-no) and maps can be kept on disk with BusterAnoFourierKeepMap=yes (default is to keep only the analysis while removing the map file). * Added support for calculation and (optional) analysis of an F(early)-F(late) difference Fourier map: * this can be selected with BusterEarlyLate=yes (default=no) and results in a MTZ file (refine_early-late.mtz) containing columns F_early-late/PHI_early-late for computation of such a map e.g. within Coot. * to perform some automatic analysis, the BusterEarlyLateAnalyse parameter can be set (default = same value as BusterEarlyLate); * When counting non-hydrogen atoms for the REMARK 3 header, count alternate conformations only once (reported by S. Sheriff). * Fixed dictionary for HEC heme. * Added modified RNA as defined at http://modomics.genesilico.pl/ (Machnicka, MA et al (2013). Nucleic Acids Res 41, D262.) if they occur in at least 10 PDB entries (as of 20151110) to list of known nucleotides MakeLINK should recognise as being covalently connected. * Fix in visualise-geometry-coot: added support for contacts across symmetry in order to give correct residue name and text in contact tab. * Some cleanup in files like LIST.html, gelly_sanity_test.log or screen_final.txt: * To avoid excessively verbose logfiles (especially for large structures), a default number of the first 100 items is now reported. For debugging purposes this parameter can be changed via the -report_numtop flag. * Some small format changes for better readability throughout. * Speedup at different stages, especially noticeable with very large structures. * Added check for PDB-style symmetry operators: if these don't fit the format NNNMMM it is most likely due to molecules being positioned very far away from the (0,0,0) origin. * Some additional consolidation in the REMARK 3 section written (follow closer the more common formatting seen in deposited PDB files). BUSTER improvements and fixes (released Nov 25 2015): * pdb2occ now reports the correct number of atoms within an altConf-based group in the comment. * Added N (generic RNA nucleotide) in order for MakeLINK to create correct linkage when encountering this residue. * Ensure our REMARK 3 records are following the most common formatting regarding text and precision of numerical values as present in deposited PDB files. Please note that for most sections of the REMARK 3 record investigated, no a single format and precision was seen after analysing all deposited X-Ray PDB structures as of July 2015. The user can adjust the required precision for each numerical (floating point) value by (a) finding the parameter name (__SOMETHING__) in $BDG_home/database/files/Harvest/header.pdb (b) setting the parameter autoBUSTER_Remark3Precision_SOMETHING to the number of significant digits one wants; e.g. refine autoBUSTER_Remark3Precision_RFREE=4 ... would change it from the default of 3 and ensure that the overall Rfree is printed as 0.2133 instead of 0.213. * Fixed problem in counting the number of hetero atoms for the REMARK 3 record (thanks to S. Sheriff for pointing this out). * Fixed problem in grade: it generated too many unnecessary pseudochirals when using MOL2 files (reported by Y. Wang). Pipedream improvements (released Nov 25 2015): * Changes to limited MR module required to ensure compatibility with Phaser 2.6.0 and CCP4 7.0. These changes have no impact on function and are backward-compatible with earlier versions of Phaser (back to version 2.5.6). * NOTE: The result of this update effectively means that the minimum version of CCP4 required by BUSTER increases from 6.3 to 6.4. BUSTER improvements (released Sep 23 2015): * Fixed bug in MakeLINK regarding symmetry description on SSBOND cards: now use full width of 6 characters (instead of only 4) as defined in http://www.wwpdb.org/documentation/file-format-content/format33/sect6.html#SSBOND. * Improved MakeLINK to avoid error about (unsupported) TAB characters in PDB file records that are actually not used within MakeLINK. buster-report improvements (released Aug 12 2015): * PDF page produced now has the same ligand information as the HTML version, including electron density picture and Mogul outliers 2D schematics and tables. * report.xml now contains all the numerical information about each ligand including Mogul outliers. * New option (-EDcontours) to set the contour levels used in the electron density around the ligand pictures. * Improve the captions for Mogul schematics by getting rid of the colored text previously used: instead describe the colors. * Ligand pages now have help hyperlinks in each section. These link to the relevant part of much improved BUSTER wiki help page. * Support for ligand corporate ID's improved with option to allow multiple ligand corporate id substitution patterns. The documentation for this feature has been much improved. * Improved the directory structure and filenames used for ligands. * Corrected bug where empty ligand pages could be produced for structures containing no ligands but monoatomic ions. * Improved the list of "boring" ligands. * Corrected bug where multiple simultaneous buster-report runs could interfere with one another by using the same temporary filenames. * Numerous other bug fixes. BUSTER improvements (released Aug 12 2015): * added modified amino-acid residues ALY (N(6)-acetyl-L-lysine, 132 PDB entries) and M3L (N-trimethyl-L-lysine, 167 PDB entries) to list of automatically recognised peptide-linkages. BUSTER improvements (released Jul 15 2015): * pdbchk_element: update to handle components.cif as of 20150626. * Added support for AddMissingSsbondRecords parameter to recover lost SSBOND records at the beginning of refinement: * Until now, as we were using MakeLINK by default for the generation of the TNT-sequence file, setting this parameter to "yes" had no effect unless "pdb2seq" had been chosen as an alternative to MakeLINK (by setting SequenceFileGeneration=pdb2seq). * This has now been changed (in response to popular demand) so that specifying AddMissingSsbondRecords=yes will automatically generate SSBOND cards for cysteine disulphide bridges that are (1) missing in the PDB file header and (2) supported by the geometry in the input model (with respect to bond distances, bond and torsion angles). This is intended for situations where e.g. a PDB model was refined with correct SSBOND records but subsequently processed in such a way that the relevant SSBOND records got lost (e.g. stripped by a MR program or step). This option is not intended as a method for inferring the possibility of generic disulfide linkages where the relevant geometry is too far away from ideal. * Setting AddMissingSsbondRecords=yes (the default is still "no") will activate the PdbChk_MissingSsbond check within 'pdbchk' (which is off by default). * The 'pdbchk' step will then use the new 'ana_ssbond' tool to check for missing SSBOND records, using bond distance, bond and torsion angles as criteria. The default criteria are based on an analysis of 28389 SSBOND records in 3906 PDB entries and use min/max values that contain at least 98% (99% for angle) of all deposited SSBOND records: CYS-SG ... SG-CYS distance : 1.948 ... 2.153 A CYS-CB-SG ... SG-CYS angle : 91.743 ... 117.406 degree CYS-CB-SG ... SG-CB-CYS torsion-1 : -130.243 ... -45.321 degree torsion-2 : 33.914 ... 154.916 degree * The presence of ZN or FE atoms close to CYS-SG atoms (3A cutoff) will exclude those CYS residues from further analysis. * Tests have shown that for well-refined structures, all SSBOND records are recovered in about 70% of cases. * Known issues: * CYS-CYS disulfide bonds across symmetry elements are not detected (we are working on this); * non-CYS disulfide bonds are not supported; * alternate conformations can cause SSBOND records to not be detected (we are working on this); * Fixed problem on OS X (10.7 and later): pdb2tls used an unsupported awk construct that resulted in failure to extract existing TLS descriptions from PDB file header. * Fixed problem when running 'refine' with an input TNT sequence file (-Seq argument): if this contained a CHAIN specification for a chain that had no atoms in the PDB file, the automatic generation of RB descriptions (-RB flag) was causing failure within BUSTER. Pipedream update (released Jul 15 2015): * Write "Pipedream" header to the top of the final output pdb file indicating the identity of the input pdb file on which the final output is based. Where Pipedream has been run with multiple input models, it will indicate which of the input models was selected as the best and used for full refinement and ligand fitting. * Renamed -seq option to -sequence to prevent ambiguity with -seqin options. Pipedream update (released Jun 16 2015): * corrected a bug that may affect the correct selection of the best input model where run with multiple input models. BUSTER improvements (released Jun 16 2015): * allow more control about geometry analysis (screening) inside BUSTER via the RunScreenInsideBuster parameter (default=yes) * removed -f flag (we did warn about this being deprecated): it had no meaning/function any longer * reporting of OS information now avoids duplication * handle ColumnName_FreeR_flag_allowed settings that don't follow the required formatting and automatically adjust for that * add check for correct syntax of autoBUSTER_hls setting - and also allow comma-separated list * 'rvalue' tool: widening of output format (for reflection numbers) to allow very large datasets like ribosomes to give correct formatting * added option -nostop to fetch_PDB tool: don't stop if a problem is encountered (like a missing reflection data file) General improvements and changes (released Mar 16 2015): * The need to obtain a separate licence key for each machine on which the software is to be run has been removed for academic users: a single valid licence key will now work on any number of machines. Pipedream updates (released Mar 16 2015): * restricted resolution range for rigid-body refinement to 9.0 - 4.0 A. This change increases the radius of convergence for the rigid-body refinement step and also increases its speed (confirmed testing on multiple test sets, including all of the pipedream tutorials). * Corrected bug whereby post-refinement would fail in cases where Pipedream is run in "cocktail" mode (multiple input ligands for Rhofit), and several ligands have the same residue name (three-letter code). * All messages output where Pipedream terminates with an error status are prefixed with ERROR: * Corrected bug whereby Pipedream would fail if reading input model from a read-only file system * Corrected potential undefined variable warning that may have occured when running the check indexing function. * NEW FEATURE: Pipedream can now be run with multiple input models. Several input models of the same target (same space group and cell) can be specified. Pipedream will analyse all of the input models against the experimental data and then select the model that best matches the input data by a variety of criteria. Please see the Pipedream documentation and also Pipedream tutorials on the BUSTER Wiki for a full description of the use and limitations of Pipedream with multiple models. BUSTER improvements (released Mar 16 2015): * Set fixed y-range for FOM-plots. * Extend the allowed range for a free flag value to [0-999]. * pdbchk_element: update to handle components.cif as of 20150227. * fetch_PDB: additional tests and checks to ensure the correct wavelength is put into the MTZ file. Also: - report a missing reflection file more explicitly - added warnings for weights below 0 or above 1 and phases below -360, above +360 or more than 360 degrees apart * Allow an alternative way of calculating Luzzati coordinate error estimate with the new tool 'luzzati'. This should be more accurate than the CCP4 program SFCHECK (since it uses the FC values directly out of BUSTER) and is also significantly faster. To switch back to using SFCHECK: please add the 'RunSfcheck=yes' parameter. * Changed default name of buster-report output directory to contain '-' instead of '.' (but user can override this via the BusterReportDirDelim parameter, which defaults to '-'). * Added method to ensure that some user-supplied residue names are output as HETATM and not ATOM (eg. if ligand is amino-acid): this requires the user to set the EnforceHetatmForResidues parameter to a (space-separated) list o chain and residue numbers (eg. "A301 B301"). * Added reporting of resolution limits also for each normal big cycle (not only for rigid-body cycles). * Improvements to reporting restraint dictionary details to output PDB and buster-report. * Updated distributed ANP and NAD restraint dictionaries. * Accept also DCY on SSBOND records in PDB files. BUSTER fixes (released Mar 16 2015): * Handle water update correctly if removed waters are part of a TLS group (and therefore contain also an ANISOU card in the PDB file). * Added check in 'diff_fourier' when comparing peaks against existing PDB file. * Fixed counting of hetero-atoms for REMARK 3 section when no hetero-atoms are present. * Fixed help message (to follow documentation, ie. including -report at the -h level). * Better checks regarding amplitude/sigma (F/Q) pairs in MTZ file (error out if no complete F/Q pairing could be found). Pipedream update (released Jul 26 2014): * Modified default resolution limits to use for the limited MR step. Only restrict high res limit to 3A and do not apply a low resolution cutoff. This bypasses a potential Phaser problem if the lowest resolution bin does not contain any reflections. * Improved and expanded on the termination messages generated in case of a Phaser failure. BUSTER improvements (released Jul 26 2014): * Restraint dictionaries for common sugars are now supplied as cif format files in common-compounds. Restraints for additional sugars also supplied. New restraints included for additional glycoprotein and polysaccharide links. * the 'fetch_PDB' utility now reports any PDB identifier which is skipped because the output directory (and files) already exist. It also reports the location (local directory or remote URL) from where the DB files were retrieved from. * added checks for sensible values to the -nbig/-nsmall command-line options for 'refine'. buster-report and Grade improvements (released Jul 26 2014): * Improved colors used in buster-report ligand Analysis page to be useful for color-blind users (and clearer for everyone). * Introduced support for ligand database/corporate ids. Use the new grade option -databaseid. The corporate ID will then be prominently displayed on the buster-report Ligand Analysis Page. Furthermore it is possible for the corporate ID to hyperlink to a relevant intranet or external web page. Correction (released Jun 20 2014): * added updated sugar restraint dictionary that was left out in the Jun 12 2014 release. Distribution fixes (released Jun 19 2014): * added two missing tools that where only documented but accidentally left out of the distribution: 'diff_fourier' and 'ana_diffmap_residue'. MakeTNT fix (released Jun 12 2014): * Corrected problem that prevented MakeTNT running under OSX. Pipedream updates (released Jun 12 2014): * Changed the minimum required version of Phaser to 2.5.0 from 2.2. This change will also have the effect of requiring a minimum CCP4 version of 6.3.0. * Added compatability checks to confirm that the binary architecture of the configured autoPROC and/or BUSTER binaries matches the architecture of the host OS. Pipedream will terminate with an appropriate message should the program and machine architectures differ. * Added verbose (-v) command line option. If selected, Pipedream will report its progress to standard output. BUSTER improvements (released Jun 12 2014): * Improved description of nomenclature used to describe sugar linkages in the supplied sugar restraint dictionary (sugar.dat) and the linkage description file (connectAt.dat). * Added restraints for sugar beta-L-fucose (FUL) to sugar.dat and added to linkage description file. ------------------------------------------------------------------- <-- Below are the Release Notes for the previous stable release --> ------------------------------------------------------------------- =================================================================== Changes included in BUSTER version 2.10.1 stable May 12 2014 =================================================================== #### Improvements to make BUSTER package easier to install/test: * comprehensive "follow me" installation instructions showing how to install, configure and test BUSTER on all major operating systems. These are available from the download page on www.globalphasing.com or included in each installation as $BDG_home/docs/installation/index.html * Environment variables can now be used to specify the location of external helper tools for buster-report, grade, grade_PDB_ligand and hydrogenate. Alternatively the tools will be located from the user's PATH. Environment variables can also be set to "none" to indicate that a particular tool is not available, and to turn off its use. * checkdeps tool: this script runs a series of checks that all the required and optional external helper tools work properly. It is designed so that a busy systems manager can quickly check that all BUSTER tools will work properly on a particular host. #### Rhofit now included in the academic BUSTER package: Rhofit is a tool for fitting ligands into difference density. It is able to change bond lengths and angles within the ligand, as well as rotate torsion angles, allowing it to search for correct ring conformations, including macrocycles. In addition, rhofit can also search for the correct chirality, if not known, for an input ligand (and the density is good enough). Rhofit uses the gelly geometry function to assess ligand strain and protein-ligand contacts, so its results will be compatible with further BUSTER refinement. Rhofit is designed to work well with BUSTER options for improving ligand density (-L). To help in visualizing results, we provide a tool visualise-rhofit-coot to use coot to quickly inspect the hits generated. Rhofit was originally released to Consortium users in Nov 2009. It is included as part of the academic BUSTER package from May 2014. #### buster-report now included in the academic BUSTER package. buster-report is a tool that provides clear, short reports of how a BUSTER run has proceeded, and an analysis of the final results including ligand geometry assessment from Mogul and protein geometry analysis using Molprobity. Please see: http://www.globalphasing.com/buster/wiki/index.cgi?BusterReport for more information on buster-report. buster-report was originally released to Consortium users in Dec 2011. It is included as part of the academic BUSTER package from May 2014. #### Pipedream now included in the academic BUSTER package. Pipedream is a tool designed to facilitate the use and integration of automated data processing with autoPROC (where installed), structure refinement with BUSTER together with (where requested) automated ligand fitting with Rhofit and buster-report into a high-throughput fragment/ligand screening pathway. Pipedream was developed as part of the SILVER project www.silver-europe.com (FP7-HEALTH-260644) for academic partners. It has subsequently been successfully adopted by a number of Consortium users. It was initially released in Aug 2011. Pipedream is included as part of the academic BUSTER package from May 2014. #### BUSTER. Improvements: * switch to using PDB v3 atom names in all tools. * Extension to the use of the -qm method for representation of ligand geometry using a quantum chemical method to allow use of GAMESS, Gaussian or MOPAC2009 programs. See http://www.globalphasing.com/buster/wiki/index.cgi?AutobusterLigandQMhelpers * -forcefield option. This provides for using the force field to represent a ligand. The initial release supports use of the helper program provided by OpenEye (http://www.eyesopen.com/) for MMFF94s: See http://www.globalphasing.com/buster/wiki/index.cgi?AutobusterLigandForcefield * new options to 'refine' command: [[-report|-reportrm] [] to run "buster-report" at end of refinement and, optionally, remove original subdirectory. * Change to the number of threads that refine will use by default (nthreads). This is now set by finding ncpu (the number of CPUs that the operating system reports). The value of nthreads is set to min(ncpu,4) if ncpu < 24, 6 if 23 < ncpu < 63, 8 otherwise. The default can be overriden by environment variable OMP_NUM_THREADS or by refine command line argument -nthreads. See BUSTER documentation $BDG_home/docs/autobuster/manual/autoBUSTER4.html#refine_default_nthreads for full details. * Allow refinement of larger structures. Increased the maximum number of atoms from 520000 to 640000. Note: this doesn't add the capability of using split PDB files to BUSTER! It is still necessary to combine such split PDB entries into a single PDB file (taking care to preserve NCS if present etc) before running 'refine'. * visualise-geometry-coot now loads up the restraint dictionaries which have been cached by autoBUSTER. * better handling of residues with insertion codes in visualise-geometry-coot * alter geometry function to prevent unrealistic short contacts between atoms involved in hydrogen acceptor-donor bond contacts. * alter geometry function to restrain chiral atoms that flatten earlier. This results in MolProbity reporting fewer C beta deviations and fewer REMARK 500 exceptions being listed for BUSTER refined structures. * pdb2tls: - improve error-detection and conversion of TLS group info - robustify interpretation of TLS selection records * pdbchk: check (and possibly correct) that - atom names are in uppercase - residue names are right-adjusted - there are no bogus oxygen charges for ASP and GLU * pdbchk_element: - update to components.cif as of 20140131 - detect and correct if element type is in lowercase * use TNT contacts for metal atoms in ligands containing supplied as .cif dictionary as these are checked and corrected. * alter range interpretation specification such as NOTE BUSTER_COMBINE XYZ { A|100 - A|200 } to not require that residue 100 is present in the file but instead be purely numeric. * extend utility distance restraints to be able to work on the distance between an atom and an atom across a crystallographic contact. This is particularly useful to prevent atoms on special positions or axes from drifting from the position/axis during refinement. * MakeLINK directives can now be added to TNT dictionaries for links so that the restraints can be automatically processed by refine -l. See http://www.globalphasing.com/buster/wiki/index.cgi?GradeCovalentTutorial * increase limit of reflection indices from 255 to 511. * improve reproducibility of the openMP-section of the scaling module (ALABASTER) when using more than two threads. * Update the NotHET set with PDB version 3 names for DNA such that a correct ATOM record (rather than a HETATM record, as previously) is written to the final refine.pdb file. Thanks to Mike Swan for reporting this. * samples/autobuster/refine_test.sh: this test script now reports SUCCESS or FAILURE. In addition the argument -nthreads can now be supplied. * added column names "FreeRflag" and "R-free-flags" to automatically recognized list of allowed test-set flags in the input MTZ file. * automatically insert into list of allowed test-set flags the one specified using command line option: ColumnName_FreeR_flag. * improve handling of what SFTOOLS thinks is an XPLOR RFREE column. * harvest: speedup the processing of REMARK records * new tool 'rvalue' to compute R-values of final mtz file at specified resolution range. R-values can be based on either Fc or Fx. * new tool 'fetch_PDB' included. This script will fetch both coordinates and reflection data from a local or online PDB archive (and convert the reflection data to MTZ format). Fixes: * fix bug that caused refine crash when temperature factor restraints for a similarity group are combined with the -autoncs_weight command. * fix memory access bug that caused refine to issue PETSC warnings and occasional crash when using -qm option to refine ligands using a quantum chemical representation. * during water-update: an error was triggered (wrongly) if initially added waters were completely removed in a subsequent stage - added check for this. * slight change in logic for excluding duplicate restraints (to avoid having same restraints in both the residue dictionary as well as the linkage dictionary). This should get rid of a few failures with 'something wrong when excluding XYZ restraint from ...' messages. * pickup any TLS-parameters after doing a rigid-body cycle. Thanks to Dirk Kostrewa for reporting this issue. * when all atoms of all instances of a specific residue name in the PDB file had their occupancy set to zero, 'refine' would stop with an error about getting the best residue. This has been reset to also allow residues with zero occupancy to be used for testing consistency between the input PDB file and any dictionary (user-supplied or distributed). * fixup precision of resolution range to avoid losing reflections. * eliminated some warnings going to stdout when running from perl. * regain openMP scalability lost via ghost-hydrogens. Thanks to Paul Rowland for assistance in discovering this issue. * fix extraction of previous X-ray weight from input PDB file: if recorded several times in REMARK 3 section, the first value is taken. Also: fix reporting if recorded X-ray weight is outside some very generous limits. * during water-update: an error was triggered (wrongly) if initially added waters were completely removed in a subsequent stage - added check for this. * slight change in logic for excluding duplicate restraints (to avoid having same restraints in both the residue dictionary as well as the linkage dictionary). This should get rid of a few failures with 'something wrong when excluding XYZ restraint from ...' messages. * pickup any TLS-parameters after doing a rigid-body cycle. Thanks to Direk Kostrewa for reporting this issues. * pdbchk: * work around "egrep: Regular expression too big" error on OS X Lion (10.7) * Fix "grep: repetition-operator operand invalid" error on OS X Mountain Lion ##### Grade: Improvements: * Grade now provided on the Grade Web Server http://grade.globalphasing.org This is useful for users without a CSD system licence working on non-confidential ligands. * Introduce override values for geometry around phosphates where mogul does not have an opinion, since RM1 does not handle phosphates well. The biggest improvement is for polyphosphates found in cofactors like FAD, ATP and NAD. * "Planarity vetoes" from Grade. Mogul information is now better used to ensure that torsions like the central one in biphenyl are not wrongly set planar when CSD distribution clearly shows a range of values. * set atomnames if supplied with mol2 with all atoms being named after their element type ("C", "N" etc). * Mogul search performed by Grade modified to allow small amount of generalisation for bonds and angles but not torsion angles. This allows better use of information from highly related CSD structures. * Introduce option -really_noH for ligands that do not have any hydrogen atoms. * grade_PDB_ligand has been switched to use v3 atomnames. * grade_PDB_ligand option -xdat to use experimental coordinates from LigandExpo when no ideal coordinates are available. * grade_PDB_ligand option -nonet to allow use from behind firewalls that prevent the tool getting direct web access. * recognise-compound utility added to find whether ligand is already in PDB chemical components definitions. This is run by grade on every run to provide a warning that a user's new compound is already a known PDB chemical component. * Support for CSDS 2014 release of Mogul. Detect and cleanly exit if mogul stalls on initial dependency check. This should enable quick diagnosis if mogul stall problem with NFSv3-mounted CSDS occurs. See http://www.globalphasing.com/buster/wiki/index.cgi?SoftwareMogulRelease2014NFSissues * Improve information about methods used in output CIF to include Mogul version, CSD version, use of additional corporate data and actual QM method used (for example RM1). * introduction of -shelx option to output restraints file for SHELXL. Thanks to Julian Holstein for help in developing and testing this new feature. * clean up unhelpful REMARK records in output pdb. Fixes: * Fix issue with torsions not being enforced during QM step of Grade * grade bug failing to clean up temporary directory on exit. * protection against Mogul run away process with malformed mol2 files. * mol2 files containing positively charged phosphorous atoms will now work. * mol2 files with titles with comments containing string "_ideal.pdb" will now work. ##### Dictionaries: * Dictionaries for iron sulfur clusters FES and SF4 added. These were produced manually using CSD information. * Dictionary for vitamin B12 (cobalamin) added. This work showed up serious issues in CCP4 and PDB definitions for B12 see http://www.globalphasing.com/buster/wiki/index.cgi?B12Dictionary * Improve contact distances for Be and Fe atoms based on CSD information. #### EditREFMAC changes: * Improve EditREFMAC's handling of quoted data - Quoted data values in the input refinement dictionary are now handled correctly - Data values that contain spaces or single quote (') characters are double-quoted when saved (even when not strictly required by the CIF format) - Input refinement dictionaries containing double-quotes that are not used for quoting data values are rejected by EditREFMAC (for example an atom name like HO3") - If the user has edited any data in EditREFMAC so that they contain double-quote (") characters, the user is warned when the edited data are saved that the result may not be usable. * Fix misfeature that could cause loss of user edits - If user clicks "Save Table" while actively editing the contents of a cell, the edited contents of the cell are now saved rather than the unedited contents. ################################################################### # # Below are the release notes of the previous BUSTER version 2.10.0 # stable (released 20 July 2011). # ################################################################### BUSTER 2.10.0, the production release of July 20 2011 incorporates the cumulative changes listed below for the beta releases of 20 Jul 2011, 17 Feb 2011, 11 Nov 2010, 8 Oct 2010, 30 Apr 2010, 26 Mar 2010 and 05 Nov 2009. ============================ Beta release of Jul 20 2011 ============================ ##### BUSTER and autoBUSTER toolbox: Changes: * autoBUSTER (refine command) will not use Hendrickson-Lattman coefficients by default. To enable the use of Hendrickson-Lattman coefficients during refinement, the column labels need to be specified as part of the command line with eg. autoBUSTER_hls="HLA,HLB,HLC,HLD". * We now default to using MakeLINK (instead of pdb2seq) to generate TNT sequence files. This should especially deal better with LINK cards from the supplied PDB file as well as using better restraints for glycosylation. * The automatic generation of dictionaries based on current coordinates (as a last resort) has been disabled. The user is required to provide a restraint dictionary for novel ligands. * We now supply grade CIF-format dictionaries for the most common cofactors, buffers and cryoprotectants in $BDG_home/tnt/data/common-compounds. These have all been checked to produce sensible results for existing PDB entries. The old dictionaries that were distributed in $BDG_home/tnt/msd and $BDG_home/tnt/ccp4 have been removed. * A detailed discussion covering recent changes in restraint dictionary tools and handling is available on the BUSTER wiki at http://www.globalphasing.com/buster/wiki/index.cgi?BusterLigandRestraints More details can be found at http://www.globalphasing.com/buster/wiki/ Improvements: * An added check within the 'pdbchk' tool will be performed: to deal with renamed RNA residues (Ar, Cr, Gr and Ur instead of A, C, G and U), e.g. coming from Coot. * refine: robustify checking of CRYST1 record * corr: work around some limitations in overlapmap * pdbchk_element: update to components.cif as of 20110702 * The BUSTER_TLS_SET usage has been improved: - automatically fuse multiple sets specified by curlys - detect and warn if there are multiple tokens on record * The various command-line programs from the autoBUSTER toolbox will now report the date of the release (and not the date of the last change to that particular command/module). This will make support requests much easier. * We now save a copy of all user-supplied and automatically fetched CIF dictionaries in subdirectory restraint_dictionaries Fixes: * pdbchk: - Fixed length of comparison with residue name (in pdbchk_element) - Fixed problem when removing lines left by MOLREP program - Fixes related to ordering of records following PDB standard - Protect StandardRecordOrder from hiccup with user specified pattern containing underscores * pdbcmb: keep top-remarks added by autoBUSTER * consolidate treatment of non-standard initial REMARK records with new parameter: RemoveOldNonStandardInitialRemarkSection default is "yes" * pdb2tls: - also print serial number in REFMAC mode - detect sequence specification without a chain name in the second token * add protection against dash failing on 'empty' .autoBUSTER files * fixup extracting TLS info from LIST.html when not doing gelly refinement eg. in MapOnly mode * write long space group name to MTZ file header * fix some grid sampling restrictions (A2, I2, I21, I4/m) * pkmaps and frcdis: robustify licence checking * pkmaps: - robustify treatment of erroneous input - fix spurious pkmaps crashes of the type: "unable to map point into map" Thanks to Xavier Robert for providing a test case. * mol2_to_cif: - Better handling of lower-case letters in atom names (Cl, Br) - Better handling of triple bonds - Use ASUNK for unknown arsenic - Handle six-column mol2 files without segfault * MakeLINK: - Fixup sorting residues with negative numbers - Correctly handle disulphide-bridges for vicinal CYS residues - Fix handling of GXPEPTIDE links given as LINK cards * TNT: up various static array sizes to deal with larger models and more reflections * install scripts: further robustification ##### grade * Write report on restraint sources into the output CIF file * Report the version of mogul used * improve handling of mol2 files. In particular mol2 files without charges are now correctly parsed. #### NOTES The 32-bit Linux distributions of Global Phasing's software will work on systems running relatively old kernels (versions 2.4.x). Most mainstream server and desktop Linux distributions that are currently under active maintenance are based on the newer 2.6.x kernel series. The current 32-bit release will be the last to support Linux systems based on the 2.4.x kernel. Future distributions (including patches and minor updates) of our software for all Linux systems released after this one can only be relied on to work on systems running kernels from the 2.6.x series. This is only a significant change for our 32-bit distribution: our 64-bit distributions for Linux have always required kernel version 2.6.x To check the version of the kernel on a particular system, run the following command: uname -r ============================= Beta release of 17 Feb 2011 ============================= * BUSTER reference card. This is a two page document that lists the most useful options for refine and grade. It is supplied as a pdf document to be printed double sided on to a card. ##### grade: New features: * In response to a user request, added the -bigplanes option so that the output dictionary contains large planes rather than the default where four-atom planes are produced. This option will make the planarity significantly more tightly restrained. Improvements: * mol2 coordinate files now handled by grade using its own code, rather than libcheck as previous releases. This enables better atom typing particularly avoiding problems with thiophene-type sulphur atoms. We recommend using mol2 format, with all hydrogen atoms present, or SMILES input when using grade. * torsion restraints produced by grade improved. Uses mogul information to set 3-fold staggered torsions for bonds joining two sp3 type atoms. Previously these could be fixed at a single 180 degree torsion, or be left free to rotate. * Documentation moved to the BUSTER wiki in anticipation of imminent release to general academic users. The BUSTER wiki is not password protected so documentation should be easier to find. Fixes: * can now handle the (syntactically-incorrect) CIF restraint dictionary files produced by Corina on structures without chiral centres * In response to a user request, grade now outputs ideal-coordinate PDB file with an occupancy of 1.0 rather than zero. ##### Gelly: Fixes: * bug fixed in and new features associated with -qm option (see below) ##### BUSTER and autoBUSTER toolbox: New features: * new ShortRun macros for running a short refinement after some rebuilding (eg in coot) of a previously BUSTER-refined structure. Use 'refine -M list' for a full description including a link to a page on the BUSTER wiki describing how and when the macros should be used. * the eh99_protgeo.dat protein dictionary now includes hydrogen atoms. Restraints for hydrogen atoms have been checked to yield sensible geometry. * A new tool is available: 'hydrogenate'. This allows the addition of hydrogen atoms to proteins and/or ligands. It uses the 'reduce' tool from molprobity but extends this to ligands described by REFMAC/CCP4-style CIF restraint dictionaries, such as those produced by grade. 'hydrogenate' is particularly useful for refining proteins at resolutions better than 2.0 Angstroms and for the -qm option. * fix for hydrogen atoms and zero-occupancy atoms: they are now both excluded from the bulk-solvent masking (whereas before they were only excluded if void correction or one of -L, -Lpdb or -x was used). There are indications that for fully hydrogenated proteins the maximum likehood X-ray calculation could be improved, this will be included in a future release. * added parameter AutomaticFormfactorCorrection (default=no) to adjust formfactor for each element in input PDB file automatically * new utility 'graph_autobuster_recipCC' to facilitate looking at the reciprocal-space correlation coefficient plot that BUSTER produces. This plot can reveal problems in the data collection/processing stage, such as: low resolution overloads, data anisotropy and ice ring contamination. It can also reveal inadequacies in the weighting of phase information from partial MR models in certain situations of severe incompleteness. Gerard recommends that the reciprocal-space correlation coefficient plot is routinely examined and the new utility makes this easier. More details on the BUSTER wiki. Improvements: * BUSTER wiki now includes more details, examples and discussion. * the distributed samples/autobuster directory now has a single quick to run example. * -l added better detection for the input dictionary is CIF or TNT or some other format (e.g. containing GELLY commands). You are recommended to directly use refmac-compatible CIF dictionaries (from grade or 3rd party tools) and not convert them to TNT format. * gelly_refine: added -Gelly command-line option and use the same logic as in 'refine' for the '-l' flag. * -qm option to use the weighted quantum chemical energy of a ligand as a restraint function in refining protein ligand complexes: * An initial check is now made to ensure that hydrogen atoms have been added to the ligand before QM can start. The 'hydrogenate' tool (see above) makes it easy to add hydrogen atoms to a ligand. * A GAMESS helper perl script now available to allow the use of ab initio and DFT methods using the GAMESS package, available from the Gordon research group: http://www.msg.chem.iastate.edu/gamess/ The helper can be downloaded from the BUSTER wiki and could be adapted to allow use of other quantum chemistry programs. * Use of -qm in refinement now summarized in a output pdb file REMARK. * Documentation of -qm much improved and moved to the BUSTER wiki in anticipation of imminent release of -qm option to general academic users. The BUSTER wiki is not password protected so documentation should be easier to find. Fixes: * dealing with user-supplied macros better by handling - superfluous spaces around the '=' character - CRLF line terminators * corrected problem in ana_pdbmaps that could affect resulting clusters (in finding padded PDB limits in fractional coordinates) * fix regression in pdb2cor where residues in inter-chain linkages were written at the top of the output PDB file * robustified the licence checking mechanism * improved the command line checking wrt. arguments being numbers * on request from PDBe have removed units from the REMARK 3 lines on the T, L and S tensor cards. Likewise removed the ALL ATOMS and OTHER ATOMS entries * corr: better handling of chains with negative residue numbers * mtzchk: robustify the handling of odd space group symbols * pdb2seq: further improved automatic detection of inter-chain SSBONDS * pdbchk_element: always automatically correct element type of hydrogen atoms in THR, VAL, ILE and GLN residues when borked by external programs * pdbcmb: improved the automatic ordering of records * pkmaps: fix some infrequently used interpolation options * BUSTER: deprecate use of old scaling, old optimiser and old structure factor engine/convolute mechanism * BUFFET: made obsolete by autoBUSTER toolbox. Please do let us know of any features which are not covered by autoBUSTER ============================= Beta release of 11 Nov 2010 ============================= ##### grade: Fixes: * fix to link library bug that prevented grade running on OSX * add missing package declaration (bug affected some CentOS and RedHat systems) * mogul search performed by grade now has generalisation turned off. Generalisation in mogul widens the search if few hits are found and can mean that non-relevant data is considered and erroneous planes produced. * grade dictionaries should now be correctly interpreted by refmac (names for planes now shortened to less than 8 characters). * correction to final REMARK 3 information in autoBUSTER pdb file if a grade dictionary is used ##### Gelly: None. ##### BUSTER: None. ##### autoBUSTER toolbox: Fixes: * pdb2cor: added work-around to deal with incorrectly placed TER records * added several work-arounds to deal with MMDB issues (through CCP4 programs like PDBSET). * better handling of cases where some (but not all) non-standard residues are given by a user-supplied dictionary ##### Dictionaries: None. ============================ Beta release of 8 Oct 2010 ============================ ##### grade: * Initial release of this new ligand-generation tool. It uses statistical information from the Cambridge Structural Database, along with semi-empirical quantum methods, to generate good restraint dictionaries from a variety of input formats. ##### Gelly: * Support for the use of quantum chemical functions as restraint functions for ligands for use in refinement with autoBUSTER and in QM geometry optimization with gelly_refine. ##### BUSTER: New features: * BUSTER (through autoBUSTER) can now directly use the weighted quantum chemical energy of a ligand as a restraint function in refining protein ligand complexes. By default the method uses semi-empirical QM methods such as RM1 with helper executable using the fdynamo program library developed by Prof. Martin Field and co-workers (http://www.pdynamo.org/). The fdynamo helper is supplied with BUSTER. Use 'refine -h' to see the options which are used, this leads you to a wiki page describing how this technique can be applied. Fixes: * several small modifications to enable use of latest Perl 5.X versions (e.g. 5.12) ##### autoBUSTER toolbox: Improvements: * when we need to create a new set of test-set flags (FreeR_flag), allow the user to override default parameters: Mtztools_FreeRflag_FrcMin=0.05, Mtztools_FreeRflag_FrcMax=0.10 and Mtztools_FreeRflag_NumAim=1000 * new parameter BusterFreeFlagValue (default=0) to explicitly set the FREFLG value for BUSTER * added check to catch inconsistent spacegroup name/number (mostly if the number doesn't follow the CCP4 N+{1000,2000,3000,etc} system * make the non-standard top header in the final refine.pdb customisable with parameters NonStandardInitialRemarkSectionCard=REMARK and NonStandardInitialRemarkSectionDelimiter=yes * also write resolution range after reporintg final/best R and Rfree on standard output. * pdb2seq: more robust handling of SSBOND cards - take symmetry information into account to decide if we should have SS linkages on CHAIN or RESIDUE cards * allow several dictionaries to contain the same residue name when testing for atom name mismatch: only if the last one also shows mismatch will we try and generate a new entry via PDB2TNT - otherwise we take the first dictionary that showed no mismatches * pdbchk: several new checks * PdbChk_SsbondHasRecordAsStars (deal with SSBOND records containing '*' characters in record number - from molprobity/reduce at least) * PdbChk_DnaResiduesFromCoot (deal with Coot's feature of renaming DNA residues to Ad, Cd, Gd and Td instead of A, C, G and T) * check that atom names don't contain comma characters * check (and fix) if PDB file has DOS-style CRLF line terminations * new parameters MissingAtomsBfac, MissingAtomsBfacSigma (and MissingAtomsBfacFactor=1.25 plus MissingAtomsBfacSigmaFactor=0.125): defaults to Wilson B (times Factor) or (if Wilson B can't be determined) to 150.0/15.0 * added option (SequenceFileGeneration, default=pdb2seq) to use MakeLINK for generation of TNT.seq file * added new parameter ReuseGeometryFile=yes to reuse the restraints from first big cycle in subsequent cycles. * pdbchk_element: update to components.cif as of 20100820. * supersede TLSInitialiseBiso with general InitialiseBiso parameter Fixes: * refmacdict2tnt: conversion of CIF chirality records with volume sign 'negative' is now correct * better handling of allowed test-set column names in pattern matching * handle parameters where the value contains multiple spaces correctly on the command-line * Pdbset(mmdb) does not deal correctly with eg. atomic charges like: O1-. There are also other issues with pdbset, so avoid using it eg. when setting the space group in a PDB file. * added check and warning if macro filenames contain characters that will confuse the sh-syntax later on * Fixup sorting water chains to the top. * Several fixes when dealing with restraint dictionaries: we have standard dictionaries, our own supplied dictionaries, a user-supplied dictionary in various formats and sometimes we need to generate restraints on-the-fly. All this needs to be done in the correct order and with all the appropriate checks (to see if a given residue name in the PDB file means the same thing in the dictionary file etc). * fix ordering of records in final PDB file when some came initially after CRYST1 ##### Dictionaries: Fixes: * Some contact distances have been fixed - various new structures involving nucleic acids and heavy alkaline-earth metals indicated that distances like Sr-N were too pessimistic ##### MakeLINK: Improvements: * MakeLINK inserts bonds based on the elements at either end of the LINK, rather than on input geometry, for LINK cards not corresponding to well-known linkage geometries * LYS-PLP, CYS-BME, HIS-FAD geometries recognised automatically when LINK cards are present * MakeLINK handles nucleic acid chains better Fixes: * MakeLINK no longer ignores LINK records between atoms which are not in close contact. ============================= Beta release of 26 Mar 2010 ============================= ##### Gelly: Improvements and changes: * Correct handling of ideal distance contacts for hydrogen atoms involved in a hydrogen bond. Fixes: * Produce useful information when bond setup causes problems because of duplicate bond angles, in replacement of the previous message "impossible(?) error in storing angles". * Target restraints setup: improve error message for duplicate atoms in target structure. ##### BUSTER: Improvements and changes: * Add formfactor for N+1. * Allow up to 512 SSBOND records (TNT). ##### autoBUSTER toolbox: New features: * Support for refinement of ligands with hydrogen atoms. The solvent mask calculations will still ignore hydrogen atoms to avoid bias. * Additional choice of water-update routine (-M WaterUpdatePkmaps): this uses the same default program for analysing difference density peaks and existing waters, but new waters will only be placed close to existing hydrogen partners (in the macromolecule). Improvements and changes: * Much improved conversion of refmac-compatible CIF-format restraint dictionary by new tool refmacdict2tnt. In particular ideal distance atomtype is converted. This tool is used on any CIF-format restraint dictionary supplied to autoBUSTER by a '-l' argument. * Online help ('refine -h') has been improved to give available arguments in order of importance (and logical groupings). * pdb2tls: default automatic definition is now identical to the '-M TLSbasic' macro (i.e. EachMacroMolChain). * pdb2tls: improved handling of phenix.refine PDB files. * A new parameter (IncludeNonStandardInitialRemarkSectionInFinalPdb with default = yes) is provided to switch off addition of the top (non-standard) REMARK section in final output PDB file. * The defaults for KeepZeroOcc and KeepHydrogens has been changed from 'no' to 'yes'. So all hydrogen atoms, as well as atoms with an occupancy of zero, will be kept throughout the refinement. However, both types of atoms are excluded from the bulk solvent masking. This means that ligands can be refined with zero occupancy hydrogen atoms included. * The 'pdbchk' tool now also checks (and fixes) PDB files coming from the MOLREP program: those files often contain lines starting with '#' to denote separate domains or molecules. * 'pdbchk' also checks/fixes a missing newline at end of PDB file. * Macros now allow setting parameters explicitly before the last cycle (even if the exact number of BIG cycles isn't known beforehand). * pdb2occ: handle insertion code when writing the Gelly commands. * Significant improvements in documentation, with updated list of parameters and more detailed sections dealing with rigid-body refinement, TLS, NCS, ligand-chasing (-L/-Lpdb) etc. * Robustified handling of command line arguments by type checking. Fixes: * When pdb2tls is used to automatically define TLS groups, but the molecule(s) are smaller than the current default (i.e. less than 100 atoms): give warning and switch off TLS usage. * When removing files from different UNIX groups, the 'refine' command might have hung: this has been fixed. * Running a subsequent 'refine' job inside a previous output directory should now be possible. * Fixed typo when getting sc-CC in single-chain models. ##### Dictionaries: * added carefully checked restraints for heme type HEC to cofactor_geo.dat. * TNT bad contacts now include hydrogen atoms. * Ideal distance contact library now has default UNK terms for hydrogen atoms and halogens. * Fixed chiral term for HYP. * Fixed some incorrect angles for d5'PEND and 5'PEND in nuclgeo.dat. ============================= Beta release for 5 Nov 2009 ============================= ##### Gelly: new features: * gelly now ignores pseudo-chiral centres. These are CHIRAL restraints where two atoms involved are equivalent. This often happened on sulfates and phosphates and could result in very ugly "umbrella" conformations. improvements and changes: * Better handling when supplied with two atoms that are exactly on top of each nother. This now produces a sanity failure rather than a numerical error. If wanted gelly can push the two atoms apart. * Option to use protein dictionary with peptide omega restrained by a torsion angle term rather than a plane (protgeo_eh99_omega_torsion.dat) ##### BUSTER: new features: * Support for some non-standard space group settings - A2, C21, I2, I21, P2122, P2212, P22121 (not yet P21221) Requires a specially patched version of sftools, for which you are referred to the Phase IV Wiki: SoftwareReleases -> known issues (descriptions and fixes/patches) -> PatchingSftools improvements and changes: * Support for 64-bit versions of all programs on Linux * Reduce the amount of memory needed for buffers in masking step by 25% in the usual cases ie. when not using the 'missing-atoms' channel * During ncsmask step detect if folding back to unit cell is required and automatically do the folding * Establish the server-client socket in BUSTER_socketdir (defaults to '/tmp') * Support for up to two million reflections (TNT) ##### autoBUSTER toolbox: improvements and changes: * make it possible to switch off the use of Hendrickson-Lattmann coefficients (HLs) by setting autoBUSTER_hls="no" * added check to stop on different spacegroup in MTZ/PDB file * improved check for a working CCP4 installation (as part of 'mtzchk') * the supplied example script for using Coot's 'findwaters' tool has been improved: added more checks for finding the findwaters binary and a patch to deal with Coot renaming residues (e.g. RNA/DNA) * added check that Rfree doesn't go above 1.0: exit with error message in those cases * plotmtv should be on users PATH after installation (this program is used to visualise graphical results in *.mtv files) * enable setting the mean electron density of both macro-molecule and solvent: RhoMacroMol and RhoSolvent, respectively * updated syminfo.lib and symop in $BDG_home/database to the versions slated for CCP4-6.1.3. This includes the renumbering of 3004 to 5005 fixes: * fixed problem when dash is the default system shell (i.e. called when running /bin/sh): 'pdb2tls' was misbehaving under those circumstances. * fixed a bug in 'pdb2occ'that could have resulted in not all required BUSTER_COMBINE cards being written, i.e. getting different occupancies for atoms that belong to the same alternate conformation (reported by a consortium member). * set GeometryWeight_ideal=0.0 when doing rigid-body refinement (should have been already in the previous release but got overlooked) * fixed a typo that caused completeness to be calculated incorrectly and put into REMARK 3 section of final PDB file. * fixed logic in finding syminfo.lib as part of the 'pdb2dpi' tool * translate spacegroups of form "P1212" etc into "P21" etc * several fixes for detecting, handling and deducing element information in PDB files (as part of 'pdbchk' section) * fixing sorting of output PDB file to follow closer to given sequence order (chain linkages confused this) * pdb2tls has seen a variety of fixes and improvements (after analysing all PDB entries with TLS information) * fixed sequence issue when adding TLS-contributions in presence of inter-chain SS-bonds. Effect is that chains are now sorted according to the TNT-sequence file * ensure input files are handled correctly (if they e.g. contain a relative path) * make it unnecessary to specify UseNmissLastCycle=yes after having set UseNmissThroughout=yes * make it unnecessary to specify UseMaxEntLastCycle=yes after having set UseMaxEntThroughout=yes ##### Dictionaries: * improved chirality information used for common cofactors =================================================================== Beta release of 22 Jul 2009, on which the production release of July 2009 (version 1.6.0) was based =================================================================== ##### Gelly: * -autoncs now has an automated weighting scheme that takes into account the number of chains in an NCS group. ##### BUSTER: * Fix resetting of bounds for B-factors: This allows TLS refinement to run to completion much more often. * Fix the internal scaling used for the rigid-body part of TLS. * Fix a corner case in the scaling module (ALABASTER) when the data are far from absolute scale. ##### autoBUSTER toolbox: * the QuickMap macro has been renamed to MapOnly * Make gelly_refine run on Darwin. * Fix a corner case in pdbtools to prevent an infinite loop when removing atoms with zero occupancy. * pdb2occ: fix when alternate conformations are for single-atom residues (like waters or metals) * give a bit more information for the (default) final void analysis step * added check that given X-ray weight is in allowed range * added checks so that -RB/-TLS flags (which have optional arguments) don't get confused by immediately following par=value settings * give warning if a fresh test-set needs creating ============================ Beta release of 8 Jul 2009 ============================ ##### Gelly: new features: * Much improved geometry function for BUSTER refinements. In particular, non-bonded contacts are more realistically modeled using an ideal-distance contact term in place of the previously used TNT bad contacts. The ideal-distance contact term is also used for 1-4 interactions resulting in improved Ramachandran statistics. Protein Bond and bond angle terms now use the Engh & Huber EH99 restraints in place of the earlier EH (1991) values (see dictionary section). * Refinement with the improved geometry function generally results in slightly lowered Rfree, improved Rfree-Rwork gaps and much improved Molprobity scores. The use of the ideal distance contact term will also lead to more realistic restraints on short protein-ligand contacts and the prevention of eclipsed torsions within ligands. * Support for TLS (see BUSTER section). improvements and changes: * Better treatment for atoms and ions on special positions. In particular sulfate ions at special positions are automatically recognized. * Much improved information in PDB REMARK 3 section. This includes adding descriptions of what is in each dictionary used, the use of -autoncs and -target LSSR options. fixes: * TNT atom selections involving ranges such as {A|1 - A|957} are now done numerically so that gaps in the sequence will not cause an error. * Early recognition of numerical problems (NaN values) in gradient calculation. ##### BUSTER: new features: * Support for TLS refinement of an unlimited number of groups. - Generally, we have found that the use of a single TLS group per chain often gives much of the benefit that can be obtained from a complicated decomposition. However, of course, in some complexes where chains are strongly entangled, the natural TLS domains will need customised definitions. - The low-level machinery to compute structure factors and gradient components by Agarwal convolution when anisotropic B-factors (actually, anisotropic U's) are specified has been implemented in TNT including the use of openMP capabilities. The input/output and optimisation mechanisms for TLS are implemented in GELLY. - An easy method to get TLS refinement going is to use the macro TLSbasic * Compute overall CC(Fo,Fc) for work, free and total. At each iteration this is written to: shell.01/reciprocal_CC.txt. * Collect overall values of the Log Likelihood for total, work and free reflections at each interation. This is written to: shell.01/mxl.txt. * Support for running on Darwin (Intel). * Support for computing structure factors and gradients when some atoms are anisotropic. * Support for more atoms (up to 200 000) improvements and changes: * Release memory for maps before entering gelly. * Set range of CC-plots to just below zero and just above one for easy comparison. * New binning routine. * Increase range for a free flag from [0-49] to [0-99]. * Allow for setting ranges for scaling parameters (see -CS command-line flag for 'refine' command) * Update to newer PETSC (2.3.2.p10) and TAO (1.8.2) libraries. * Link against CCP4 version 6.1.1 (patched). fixes: * Compute R-values on observational scale. * Fixup finding the TNT symmetry information file, e.g. for "P 1 21 1" when running through Buffet. * Fixup generation of R_Rfree.html if there are empty bins. * Remove support for PlotIt13 ##### Dictionaries: new features: * Engh & Huber (1999) restraints are used for proteins; in particular, cis- and trans-proline are now distinguished. * Torsion restraints for proteins have been very carefully checked and are now active by default. Proteins and nucleic acids have atom-typing which, in conjunction with the new ideal-distance contact term, provides a 1-4 term making many implausible conformations strongly disfavoured in optimisation. * The dictionary format now allows the inclusion of ideal-distance contact information via atom typing, and this information has been included for proteins, nucleic acids and commonly-found ligands. improvements and changes: * Sugars now have enforced chirality agreeing with the PDB; there are optional chiral linkages to enforce biologically-correct chirality of glycosylation. * Some common buffer molecules have curated geometry restraints and atom typing; we now agree with the PDB as regards the names and geometries of the many variants of DTT. fixes: * Corrected some missing angle restraints on nucleic acids and protein termini * Leaving-atoms added to sugar descriptions, so restraints are now correct for both polysaccharides and isolated sugars * Improved descriptions of chiralities of cofactors * Azide is now restrained as linear rather than bent, and some quaternary ammonium geometries made more plausible. * Introduced new BCORREL terms for haem based on unrestrained re-refinement of very high resolution structures * MakeLINK now handles RNA more correctly, recognising some more of the common modified bases. ##### autoBUSTER toolbox: new features: * New tool visualise-geometry-coot for allowing users to run efficiently through lists of geometry problems in the context of the structure and electron density; this is a very useful counterpart to the list of difference-density peaks already available from coot. * pdbchk will check (and optionally fix) the position of an OXT atom within a residue: for the internal conversion routine this needs to be at the end of the residue * warning messages from Gelly (during initial sanity check) are now presented in main logfile (i.e. stdout) * command-line flag (-w ) for easier setting of initial X-ray weight * user can explicitly specify a set of Hendrickson-Lattmann coefficients using the autoBUSTER_hls parameter * command-line flag -nthreads (to specify number of CPUs/cores to use) * macro feature: see 'refine -M list' for a list of pre-defined macros and documentaiton about concept and user-defined macros * support for TLS refinement (-TLS flag) * tool to extract TLS information from PDB file header or to generate simple TLS parametrisation: pdb2tls * tool to generate template for occupancy refinement (alternate conformations, partially occupied residues etc): pdb2occ * tool to calculate DPI values: pdb2dpi. THese are now generated by default and placed into REMARK 3 section of final PDB file * New tool pdbvconv (now used automatically in pdbchk) for converting PDBv3 input to the PDBv2 format used internally; includes an automatically-generated complete list of atom names that changed between v2 and v3. * sourcing the $BDG_home/setup.sh or $BDG_home/setup.csh scripts (for setting up the proper environement for running BUSTER) now removes all references to older installations before setting up a clean environment. This will make it easier in the future to switch between different versions of the program. improvements and changes: * final results file (refine.mtz and refine.pdb) are now both files instead of symbolic links * report number of swapped and/or pruned residues (if RunBusterPrintUsefulInfo is set to yes): mostly of interest for refinements with LSSR_type NCS restraints. * X-ray weight now reported (and used) in correct units (instead of x100 as before) * added some safety net when running with missing atom channel active * changed B-refinement scheme defaults so that we basically only ever decide between 'individual' and 'None' * allow several column names for test-set flag - currently FreeR_flag and FREE (but expandable) * new/changed defaults: - 100 small cycles - UpdateWaters=no (i.e. -noWAT); to switch it on use -WAT flag. - low-resolution cutoff now less sudden: only kick in after big cycle 2 and restrict lowest resolution bin only partly. - BUSTER uses an improved (resolution) binning mechanism: this can give different results, but avoids using too many bins with very few reflections or a ridiculously small amount of bins for large numbers of reflections (previously this was basically fixed at a value of 9 bins). * pdbchk will now check for PDB v3 versus v2: internally autoBUSTER currently uses v2 (and final output PDB file will also be v2) * old flags -n/-N now called -nbig/-nsmall * exact command-line now printed with line-break (and \-character) at top of refine.pdb * if input PDB file contains the recorded X-ray weight, this will be used as a starting value (parameter StartFromPreviousWeightInPdb) * print anisotropic B-factor scales and Luzzati error to meaningful precision in REMARK 3 section of refine.pdb * implemented an option (CorrRunSeparateChains) to calculate CC values separately per chain instead of all chains in one go: this is required for large structures with lots of chains (since otherwise the ATMMAP runs out of integers) * possibility to use fixed solvent scales (might be necessary for very poor models or very low resolution date): FixedSolventScales parameter * added test PdbChk_IleResidueWithAtomCd to deal with PDB files coming out of phenix.refine (or CNS/X-PLOR) * added test to check CCP4 right at the start: MtzChk_WorkingCcp4 * added parameter (KeepFromEnv_ccp4) to enforce using CCP4 from current environment fixes: * resolution limits from different sources (user or MTZ file) are now synchronised so that values in REMARK 3 section of final PDB file should be correct. * fixed UseGapAsBreakInSeq feature in pdb2seq * keeping hydrogens (e.g. on ligand) now works correctly with PDB2TNT (generating dictionary frmo existing structure) * Fixup sorting with negative residue numbers in pdb2seq * renamed pdb2tnt to ab_pdb2tnt to avoid clashes with PDB2TNT on case-insensitive HFS+ filesystems * Take HETATM records into account due to MSE residues: this should support Se-MET structures better * fixed detection of system-wide .autoBUSTER file ============================= Beta release of 11 Jul 2008 ============================= ##### Gelly: new features: * NCS restraints - The use of Local Structure Similarity Restraints (LSSR) is a major new development in this release. The concept of LSSR is novel: it is not based on the 3D superposition of coordinates, but instead works by restraining corresponding local interatomic distances to be similar. A plateauing functional form is used for the restraint. LSSR are much easier to use than the conventional "soft" NCS restraints. In particular, the need to prune out regions of the structure that violate NCS is markedly reduced. In addition, because of the local nature of LSSR, domain-type motions are allowed to take place without any manual intervention. The autoBUSTER command line option -autoncs automatically identifies NCS relations and sets up the corresponding LSSR restraints. * Target restraints - LSSR can also be used to exploit pre-existing knowledge that the structure under refinement is similar to another already solved structure, called the "target". The target structure remains fixed during refinement. For instance a high-resolution apo structure can be used as a target when refining a lower-resolution ligand-soaked molecular replacement structure. Use of LSSR in this case produces improved Rfree together with better geometry validation scores. The method is also particularly applicable to the refinement of proteins with multiple crystal forms. The autoBUSTER command line option -target facilitates the easy application of target LSSR. * Procedure for swapping equivalent atoms to improve NCS and/or target restraints. * Utility restraints on the angle between any three given atoms. An example application of this is to allow the definition of restraints on the geometry of metal binding sites, such as tetrahedral zinc sites. * Utility restraints on the torsion angle between any four given atoms. The restraint function takes the form of a sinusoidal penalty. A potential application would be to ensure that an aliphatic part of a ligand is encouraged to remain staggered. improvements: * More helpfully worded messages upon failure of sanity checks. ##### BUSTER: new features: * use FFTW for P1 transforms (solvent, prior and MaxEnt) New keyword: FFTW={0,1} to dis-/en-able. Default is enabled. improvements: * improved handling of 'unknown' element type 'X '. fixes: * [cmdServer]: fixup a signal race. ##### autoBUSTER toolbox: new features: * give short summary about severe outliers before and after each BIG cycle * better automatic assignment of atom type to PDB records to use correct form factors for each atom. Columns 77-78 on output PDB file should show correct atom type. improvements and changes * can now specify which checks on PDB files should not be done in "pdbchk" (parameter PdbChk_ChecksNotToDo) * the automatic restiction of poor low-resolution can now be controlled using parameters AutomaticRestrictLowres, AutomaticRestrictLowresCcCut, AutomaticRestrictLowresBinCut, AutomaticRestrictLowresFromCycle. The current defaults make autoBUSTER behave as before. * the command-line flags -Edat, -f, -t and -Ncs are now deprecated (but still recognised). Using -Gelly instead should give the same results. * additional command-line flags added that are passed directly to Gelly: -autoncs, -target:f, -autoncs_weight, -target_weight, -sim_swap_equiv, -sim_swap_equiv_plus, -glim, -screen:i, -screen_sigma, -autoncs_noprune, -dlim, -keeppdb See Gelly documentation for details. * several -Gelly cards can now be given on the command-line. * show real-space CC plots always in range 0.0 - 1.0 * output basic REMARK 3 information about NCS and TARGET restraints * the values for selecting the 'interesting region' (during ligand detection with the -L and -Lpdb argument) are now under parameter control: AnaPdbmapsMinVol, AnaPdbmapsPadding, AnaPdbmapsCut1, AnaPdbmapsCut2 * detection of 'interesting regions' now much faster * new flag -noNORM to prep_FitMAP (to supress normalisation of maps). * update to handle PDBv3 components * check that PDB files and MTZ file are of this type (in case of giving the wrong file name for a specific flag) fixes: * updated link to current PDB format * the logic for the formfactor correction was wrong (taking CuKa values as a base-line): this has been corrected. * better dealing with atom or residue names that contain '*' character * when checking special linkage descriptions in a (user-supplied) TNT sequence file: deal with continuation lines and linkages specified on CHAIN record (i.e. inter-chain linkages). * make sure final MTZ file is in standard CCP4 asymmetric unit * generation of cluster.lis by prep_FitMAP had problems when encountering more than 10 regions. * all scripts: improved handling of UN*X utilities: head, tail, sort, echo on different distributions and platforms. ============================= Beta release of 15 Feb 2007 ============================= new tools/programs: - pdbchk: new program for extensive checking of a PDB file. In 'fixing mode' (i.e. if an optional output file name is given) it tries to correct some of the more trivial problems it encounters. The checks are for conformance to the PDB format specification (up to a point), but also for some autoBUSTER-specific requirements (like missing chain identifiers). The goal is to ensure maximum compatibility of the PDB file with the autoBUSTER refinement toolchain. - pdbcmb: will combine two PDB files (old and new) to produce an output PDB file that will contain everything from new PDB file plus any PDB header records only present in old PDB file. This is mostly used internally to carry through all header records of input PDB file to the final refine.pdb file. - mtzchk: tool for checking MTZ file (cell parameters, space group, amplitudes and sigmas, ...) new features: - The user can override the default scratch directory (including a file-prefix) by setting the environmental variable autoBUSTER_scr. However, since the names of temporary files are now ensured to be unique (and no name clashes between different autoBUSTER jobs should occur), this will be rarely necessary (if at all). - three levels of adjusting defaults are now possible (via parameter setting in a file called .autoBUSTER): a) installation-wide in $BDG_home/autoBUSTER/bin/linux/.autoBUSTER b) user-wide in ~/.autoBUSTER c) current directory in ./.autoBUSTER - The automatic generation of a TNT sequence file (via the 'pdb2seq' command) has been improved by adding the following features: * breaks in consecutively numbered residues are detected (based on C -> +N bond length for amino-acids and O3* -> +P bond length for RNA/DNA) * connections of non-consecutively numbered residues are detected (based on C -> +N bond length for amino-acids and O3* -> +P bond length for RNA/DNA) * chains with several OXT atoms (i.e. C-terminal oxygens) are handled * detects potential modified amino-acids (residues that contain all atoms of a peptide group) * better handling of RNA/DNA models * LINK records in PDB file are now used (by default), if the atoms referred exist and the described connectivity is within a sensible bonding distance. If the two residues named in the LINK record are already connected by a standard linkage (like PEPTIDE or SUGPHOS), no additional link will be created. Also: if at least one of the atoms referenced is a metal no restraints will be generated. * inter-chain SSBOND records in input PDB file are now supported (generates SS linkages on the CHAIN card of the TNT sequence file) - A hook is provided for a user-supplied water updating routine (which will then be called instead of the built-in procedures that use PKMAPS or ARP_WATERS). improvements: - The calculation of real-space correlation values (with the 'corr' command) handles fragmented chains better. It should also not be thrown off when the order of chains and residue numbers is unusual/unexpected. - The final PDB file will carry over all relevant header records from input PDB file. The REMARK 3 header (refinement statistics) is obviously updated. - The main 'refine' command will now report the list of command line arguments to standard output. - The used set of standard dictionaries can now be changed under user-control. - The order in which restraints for non-standard residues are generated can be defined by user. By default, the dictionary is searched for in order: $BDG_home/tnt/data/msd, $BDG_home/tnt/data/ccp4, MakeTNT. - A non-standard residue for which there is a distributed dictionary available will be checked to make sure that the PDB file uses the same atom names as the distributed dictionary file. If this is not the case: a new dictionary will be generated on-the-fly (from coordinates of the best representative in the PDB file) using the program PDB2TNT from the MakeTNT toolbox. However, if the dictionary in error is a user-supplied one (given with the -l flag to the 'refine' command), an error occurs and the user is reminded to fix that external dictionary file (or the PDB file). - pdb2tnt will detect (and remove) duplicated restraints for non-standard residues (which have restraints generated on-the-fly from the PDB coordinates and are connected to other parts of the structure via some standard links, e.g. a PEPTIDE bond). - The new parameter KeepCurrentRmsBond (default = no) can be set to 'yes' if autoBUSTER should adjust the weight between X-ray and geometric term so that the final rmsd(bond) stays the same as the value for the input PDB file. - Gelly is used for some initial sanity checks: * detection of unusual short contacts (which can point to problems in restraints, TNT sequence file or PDB file). * detection of duplicated restraints (shows problem in restraints dictionary) * reporting of missing atom-atom contact distances - Generally much more initial checks at the start of an autoBUSTER job with (hopefully) meaningful feedback to the user if problems are encountered. - After the PDB file has been (internally) converted to ATOMC format (*.cor file), the atom types encoded in that new file are checked against the formfactor database. This will detect problems when translating a PDB file: * no or bogus atom type in columns 77-78 of PDB file * atom name (columns 13-16) in PDB file can't be used correctly to deduce atom type - The handling of atom and residue name for waters has been generalized and now follows the PDB standard: water atoms are named " O " and residues "HOH". - the two parameters FinalMapsNormalized (default=yes) and FinalMapsCoverPdb (default=no) can be used to specify how the 2Fo-Fc.map and Fo-Fc.map file should be generated. fixes: - Now deals better with residue names (in PDB files) that are less than 3 characters long. - The collecting of all required geometric restraints (from distributed standard dictionaries, user-supplied dictionaries or restraints created automatically on-the-fly inside autoBUSTER) now also knows about link types in addition to residues. This allows complete control of unusual linkages within the macro-molecule (attached sugars, modified amino-acids, covalently attached ligands etc). - When the starting structure shows very bad geometry (based on the value of rms(bond)): only run structure idealisation if in BUSTER-TNT mode (as opposed to the default BUSTER-GELLY-TNT mode). If structure idealisation is needed, it is highly recommended to use Gelly directly (via the 'gelly_refine' command). - Internally in autoBUSTER, the SCALE cards of a PDB file will always be removed. This is due to the lack of precision when going from orthogonal coordinates in a PDB file to the ATOMC format or *.cor file used by BUSTER, TNT and Gelly internally). The effect of not removing the SCALE cards was a slight coordinate shift - even if no positional refinement was done at all. - The program 'cor2pdb' has seen several fixes: however, since the Gelly part of BUSTER now writes a PDB file directly this tool is not used by default anymore. - When a TNT restraints dictionary needs to be generated on-the-fly (using PDB2TNT from the MakeTNT toolbox), the extraction of the best residue out of the user-supplied PDB file has been improved. For different copies of the same residue/compound it takes into account * different levels of completeness * B-factor * occupancy * alternate conformations The best occurrence of a specific residue is the one with the lowest B-factor which at the same time is the most complete. - Residues are no longer renumbered, since all programs can now deal with negative residue numbers (e.g. if a His-tag is present). ============================= Beta release of 31 Aug 2006 ============================= new features & changes: - 'refine' will now check if an autoBUSTER job was already run in same directory (or subdirectory as given with -d flag): this will trigger an error. - real-space correlations between final model and maps (using the 'corr' tool) are now calculated using the full FC/PHIC as available from BUSTER (i.e. including bulk-solvent correction). Before, only the structure factors from the actual coordinates were used. - user-defined dictionaries (-l flag to 'refine') now also accepts CCP4-formatted CIF files (which get converted internally to TNT-formatted dictionaries on-the-fly). This should make it easier to use autoBUSTER without the need of converting already available dictionaries. new tools/programs: - new tool prep_FitMAP to prepare all files necessary for running M. Brandl's ligand-fitting program (FitMAP, part of the MakeTNT toolbox) in the correct way. fixes & improvements: - improvements to the program that analyses maps and model for 'interesting' regions (ANA_PDBMAPS). This might make it easier to use it for other tasks as well. - more control on convergence when adding waters: parameter KeepAddingWatersAfterN (default=yes) will increase the total number of BIG cycles to be done as long as water update still occurs. - added 'weight_ncs' to the available settings in 'gelly_refine' - fixed a nasty typo in 'refine' (showed up on SuSE 10.1 only so far) - R/Rfree values in analyse.html are now correct again - fix in 'gelly_refine' that caused problems with GLY residues - fixes within the underlying logic of autoBUSTER refinement jobs that make it possible to run on machines even if these weren't configured as MASTERs ============================= Beta release of 13 Jul 2006 ============================= new features & changes: - uses BUSTER-GELLY-TNT, i.e. new optimiser and new geometric restraint handling - new program (ana_pdbmaps) to analyse density maps for potential ligand binding sites ('interesting' regions). This doesn't require the present of 'water' atoms anymore. This is now the default. - the minimum distance to existing atoms for placing new waters has been changed from 1.5 to 2.5 A (due to using ana_pdbmaps - see above) new tools/programs: dock_screen: initial version for general FFT-based docking, giving starting points for further refinement via torsion angles pdb2rig : simple tool to create rigid-body definitions templates seq2seq : simple script to convert ASCII sequence into TNT sequence file pdb2cor, cor2pdb: simple tools to go from/to ATOMC format gelly_refine : script to run gelly standalone, i.e. geometry-only refinement mk_o_macros.sh, mk_coot_macros.sh, mk_pymol_macros.sh: to generate scripts for graphics programs O, Coot and Pymol documentation: - up-to-date version with more examples and documentation about important flags/parameters - added documentation for additional tools - updated and enhanced documentation for 'refine' command - 'refine -h' is updated and also points to more extensive HTML documentation fixes & improvements: - handling of atom names with spaces (converted to underscore) - already present PDB header records are carried through - deal with residue names that are shorter than three characters (left or right justified) - the flag '-noWAT' can also be given an (optional) number of BIG cycles for which it should be used (e.g. skipping water-updating only for the first 3 BIG cycles) - some advanced settings can now also be given on the command line (using a parameter=value syntax) instead of placing them into a .autoBUSTER file. - added checks to avoid using FOSC/SIGFOSC and HLA-D if a MTZ file is used that is actually the output file of a previous BUSTER or autoBUSTER run - if the input PDB file needs addition of a chain ID: this is now also added to SSBOND, HELIX and STRAND cards present in the input PDB file - rigid-body defining files (given with the -RB flag) can have either the autoBUSTER-specific format or use directly the advanced GELLY-syntax - several fixes in 'pdb2seq' to have a correct output sequence file even in unusual cases (dealing with C-terminal oxygens, gaps and breaks) - no need for running 'newmachine' for every single machine the program should be run on. If the first machine the program is installed is e.g. of type Linux, then all further machines of the same type will fetch this initial configuration. So only if machines of the same type have a different setup (in respect to crystallographic software, scratch space, very different OS etc) the 'newmachine' command needs to be used. - only interesting regions of at least 50 A^3 volume are analysed (when giving the -L flag) - several fixes to allow user-control over B-factor refinement (see -B flag) - occupancy refinement should work now =================================================================== Any feedback and remarks should be send to the BUSTER developers: buster-develop@globalphasing.com ===================================================================