|Home||About CCP4||CCP4 Projects||Downloads||Documentation||Courses||Developers||CCP4 people||WG1/WG2|
Note that there are several independent documents linked here covering different platforms, different versions of CCP4 and different components of CCP4. If you can't find your problem here, please email us.
A list of more general version-unspecific problems can be found on the General Problems Page as well as descriptions of various operating system-specific eccentricities.
There are also a number of platform-specific pages:
A list of general and version-independent problems with CCP4i can be found here:
Note that an issue exists with WinCoot 0.6.2, previously distributed by CCP4. In this version the program findwaters crashes midway through its run. This program can be solved by replacing findwaters-real.exe in WinCoot\bin with the version available from Bernhard's site
WinCoot 0.6.2.1 is unaffected by this problem.
Phaser fails to search in the selected alternative spacegroup when one alternative SG is selected.
The workaround is to either modify the space group in the MTZ header or select more than one alternative spacegroup to search in. Fixed executables and source code will be available in the near future.
More details are available from the Phaser wiki bugs page.
Libcheck produces a garbled model for a SMILES string in certain cases. This will be observed in any program that calls libcheck to generate restraints information such as Coot or Jligand.
The current workaround is to replace the CCP4 6.3.0 version of libcheck in the $CCP4/bin directory with the version from the previous CCP4 release (6.2.0). These can be downloaded for Linux, MAC and Window from here:
An fixed version for the current libcheck will be available in the near future.
(Thanks to Hari Jayaram)
The beta version of AMPLE distributed with the original CCP4 6.3.0 (downloaded from CCP4 site before 15/08/12) has several minor bugs. It may complain of a missing argparse module.
An updated tar-ball is available (below) which should be extracted from the top level CCP4 directory ($CCP4). It will replace older AMPLE files and correct the problems.
The updated tar-ball is available from here: ample-update.tar.gz
Areaimol, when compiled with certain compilers (e.g. gcc 4.4.3), will give a segmentation fault or give garbage values for its calculations (NAN's and Infinities). This was caused by a memory leak in the code.
The source code has been fixed and the memory leak removed. Updated binaries for Linux systems (32 & 64 bit) and Windows are available below.
(Thanks to Phil Evans)
cprodrg, when run on pdb files with HETATM records fails after running refmac
The source code has been fixed. Updated binaries for Linux systems (32 & 64 bit) and OS X are available below. For source code compilation, the c_refmac.f file should be replaced by the new one.
(Thanks to Andrew Sharff)
The hgen interface fails to open with an error in "RunTask".
A replacement hgen.tcl is available.
When running any of the cctbx dispatchers (phaser.sculptor, phaser.ensembler) or xia2 you may see the following error message:
ImportError: __import__("boost_python_meta_ext"): /usr/lib64/libstdc++.so.6: version `GLIBCXX_3.4.9' not found (required by ...../ccp4-6.3.0/src/phaser/phaser-2.5.0/build/intel-linux/lib/boost_python_meta_ext.so)
This is most likely due to having a local gcc installation that is too old to support GLIBCXX 3.4.9.
A quick fix is to place the following libary in the phaser build lib directory ($CCP4/src/phaser/phaser-2.5.0/build/intel-linux/lib/):
The dispatchers should take this version in preference to the system version whilst not effecting any other components on the system
pdb_extract build fails on ubuntu as the file platform.sh contains bash specific code. The workaround is to replace $CCP4/src/harvest_app_/pdb_extract/etc/platform.sh by updated file.
(Thanks to Clemens Vonrhein)
On systems without the file tcl.h in the header include path, the CCP4 build script fails to build the library $CCP4_LIB/libDiffractionImage.so which provides the Tcl interface to DiffractionImage. As a result idiffdisp fails to run.
If this file is missing, rebuild DiffractionImage explicitly specifying the location of the tcl.h header. This can be done by modifying the configure command logged near the top of $CCP4_LIB/DiffractionImage/config.log. For instance, if you have the CCP4 tcltkplusplus bundle unpacked under $CCP4_MASTER, then run the configure command in $CCP4_LIB/DiffractionImage/, with the same options listed in config.log but also appending
CXXFLAGS=-I$CCP4_MASTER/tcltkplusplus/includeFollowing this, run
make make installNow idiffdisp should run correctly.
There is a serious error in all the 1.5.x versions of Pointless for certain cases, fixed now in version 1.6.1 available now from our ftp site. The wrong 1.5.x versions have been distributed by CCP4 in releases 6.1.24 and the recent 6.2.0, but earlier CCP4 releases (6.1.13 and before) contained earlier versions of Pointless (eg 1.4.x versions) which do not suffer from this bug. However I know that people have installed 1.5.x versions from our ftp site.
I would like to apologise for problems caused by this error
=== Corrected versions
These replace the distributed versions
=== What cases does the bug affect?
The bug may scramble the order of symmetry operators relative to the ISYM flags on each observation (see below, technicalities), leading to an incorrect reconstruction of the original hkl indices in Scala or Aimless. This happens for MTZ input files (usually from Mosflm) integrated is the highest point group in higher symmetry lattices, specifically tetragonal 422, hexagonal 622, rhombohedral R32 and cubic 432, when the output file is also in the same point group. In these cases, the output file will be wrong even if Pointless is just copying from input to output ("-c" or COPY options)
The bug does NOT affect:
=== How serious is it?
Files with scrambled symmetry operators will lead to wrong reconstruction of the original indices which are used in Scala or Aimless to generate beam directions for the SECONDARY absorption correction. Thus the error will have the biggest effect on long wavelength data with substantial absorption, and have much less effect on typical short wavelength data. High multiplicity will also tend to average out the errors.
Unmerged MTZ files store for each observation the hkl indices reduced to the asymmetric unit and a flag ISYM which is the index into the list of space group symmetry operators stored in the file. This information can then be used to reconstruct the original hkl indices measured. This bug caused to order of the symmetry operators to be changed without changing the ISYM flags, leading to a wrong reconstruction of original hkl in Scala/Aimless, and hence wrong secondary beam directions for the absorption correction.
(Thanks to Phil Evans)
The input fixed model mode of molrep fails with the message (linux version)
Fortran runtime error: End of record
The XDS XSCALE output from pointless causes problems in reindexas used by xia2.
These replace the distributed versions
Certain jobs fail occasionally reporting Arithmetic exception when the distributed 32 bit phaser binary is run under 64 bit Linux.
A fixed version is available from the ftp site. This version also replaces the distributed binary in automated downloads from this point onwards.
Sfcheck fails for a large mtz-file with the message (mac version)
forrtl: severe (174): SIGSEGV, segmentation fault occurred
crank will fail if peaks are found in the difference map. Replace ccp4i/crank/plugins/bp3/crank_bp3.tcl by the updated file.
(Thanks to Raj Pannu)
phaser MR packing test cannot be turned off. Replace ccp4i/tasks/phaser_MR.tcl by the updated file.
(Thanks to Airlie McCoy)
phaser MR The incorrect keyword NPEAKS is written to the command script rather than NUMBER. Replace ccp4i/tasks/phaser_MR.tcl by the updated file.
(Thanks to Martyn Winn)
View files in coot The dictionary files are not transfered to coot. Replace ccp4i/src/database.tcl by the updated file.
To view output of phaser_MR in coot(if available) replace ccp4i/src/plugins.tcl by the updated file.
Edit Restraints in PDB File. The leading space for one-letter atom names was lost when reading LINK records from PDB files. Replace ccp4i/utils/pdb_utils.tcl by the updated file.
ProDrg - restraints preparation. The cprodrg executable was not recognising the title keyword. Replace ccp4i/templates/prodrg.com by the updated file.
The His CD2:CG bond length and Trp CD2:CE2 bond length differ from values published by Engh and Huber.
We recommend updating to the latest ligand dictionary, available here
(Thanks to Cho Yeow and Mitch Miller)
Description of problem Missing data in the tables used for the truncate procedure on centric reflections can cause these to erronously be set to 0.
Replacement files for ctruncate are available for various platforms and the source bundle. These should replace the existing ones.
The bundles for linux and i386 OS X have also been updated.
(Thanks to Jeff Headd)
Buccaneer and refmac do not cycle as expected.
A patch is provided that can be untarred over an exisiting distribution.
(Thanks to Kevin Cowtan)
ArpWarp fails to install
Dear Colleagues, some of you might have noticed that ARP/wARP does not install cleanly with the latest version of CCP4, 6.1.13. We did some experiments ourselves and can confirm the observation of an incomplete installation that is only partly functional, the lack of the possibility of manual deinstallation and the following error message that appears at the time of running the ARP/wARP 'install.sh' script: can't read "getcontentlist_array(TASK_MODULE,206)": no such element in array NOTE: You can skip the next paragraph if you're not interested in how I reproduced the error and proceed to a bug fix. Sorry for the long paragraph. Two scenarios of CCP4+ARP/wARP installation were tested: Mac OSX 10.5 and Ubuntu Linux 9, x86_64. Both the latest versions of both softwares. The download of CCP4 was made from ccp4.ac.uk, a typical installation chosen, which included also the tclsh/bltwish components that the GUI depends on. The dmg-installer was used on the Mac to install into the folder /Applications, on Linux the downloaded file was just untarred in a folder of choice followed by running CCP4's 'install.sh'. The 'ccp4.setup_sh/csh' file was placed into the user's shell config file by sourcing it there: '.bashrc', '.cshrc', '.tcshrc' depending on the shell used. Starting from a terminal that has the new CCP4 environment setup, the ARP/wARP installation could be done. On the Mac however it is more complicated because the superuser owns the installation. Through 'sudo xterm' and in there 'source ~user/.cshrc' (to be replaced by your settings) the ccp4 environment is setup for the superuser. In both cases (on the Mac as superuser) 'cd' to the ARP/wARP folder, e.g. arp_warp_7.1, and run ARP/wARP's 'install.sh'. This script tries to install ARP/wARP into the CCP4 GUI automatically and this is where problems start. The error message from above, the 'Program List' contains no buttons from the installation, these are just in 'Model Building', manual deinstallation is not possible, the new tasks don't seem to be registered with the CCP4 GUI. The fix: The tcl-files that control the installation of a new software module read the file '$CCP4I_TOP/etc/UNIX/modules.def'. This file seems to have a line missing: In the block of 'TASK_MODULE' lines (around line 500) the entry 'TASK_MODULE,206 programlist' is missing. Please add it there at about the right place. Rerun the ARP/wARP installation and the error message should be gone, the installation should be complete, i.e. ARP/wARP buttons in 'program list', too, and the possibility of manual deinstallation. I hope this is helpful. Cheers, Gerrit.
(Thanks to Gerrit Langer)
No output from bplot stage of procheck.
A replacement file is provided that replaces the check.script in ccp4i/scripts.
This is a bug in the Pisa code that looks at the environment settings for the program. Running through CCP4i it all works fine but outside CCP4i it causes a segmentation fault when it can't find a particular environment setting that it requires.
A fixed Linux executable is available: pisa
A missing $ on line 117 of crank_shelxd.tcl causes a test to be incorrectly interpreted.
The fix is to replace the $CCP4I_TOP/crank/plugins/shelxd/crank_shelxd.tcl with
sketcher fails with the error can't read "array(CHILDREN)": no such variable when attempting to edit the element type or planar groups.
The fix is to replace the $CCP4I_TOP/sketch/sketch.tcl with
When importing a file to MTZ format using f2mtz or the import feature in the GUI, if using ctruncate to generate structure factors, the output MTZ doesn't include an F or SIGF column. This is a bug where the MTZ output from ctruncate isn't copied over to the output MTZ for the job.
The fix is to replace the file $CCP4I_TOP/scripts/import.script with
(Thanks to Loes Kroon-Batenburg)
If Scala is run with a Crystal (or Project or Dataset) name containing more than twenty characters, the output mtz file will have cell dimensions set to zero and consequently be unreadable by other CCP4 programs.
The fix is to download two updated files:
to the $CCP4/src/scala_ directory and remake scala.
(Thanks to Wenjian Wang)
Under some circumstances, most likely the presence of ice rings in the data, Ctruncate may fail to give amplitude data in its output mtz file and attempting to view the Wilson plot with loggraph will result in an error.
The following patch to ctruncate.cpp provides a fix:
(Thanks to Norbert Straeter and Maria Rodriguez)
The procedure for including a file with known HA sites needs to be modified to take into account whether one is using SAD/MAD or SIR/MIR.
This can be done by using three updated ccp4i files:
(Thanks to Clemens Vonrhein)
This replacement file corrects the problem using a updated ccp4i file:
(Thanks to Liz Potterton)
mapslicer fails to launch in ccp4 linux binaries, and when built using the default build.
This is due to missing symbols from the ccp4c library. Several workarounds exist
When selecting amplitude based twin refinement in ccp4i, the option is not passed to refmac5.
This replacement file corrects the problem using an updated ccp4i file
(Thanks to Raj Pannu)
Attempting to rerun an old Refmac job results in the job getting stuck in the 'STARTING' state. On non-Windows platforms there is also an error message:
"ATOM(0)": variable isn't array ...
This is because changes have been made to the Refmac interface in CCP4-6.1.1 which are not backwardly compatible. This replacement file corrects the problem using an updated ccp4i file
Attempting to define the dataset and crystal names through the scalepack2mtz interface, or import_scaled interface does not work. A workaround and fix using replacement files is given
If the Import Scaled task is run with the "Ctruncate" and "Use dataset name as identifier to append to column labels" options selected, the job will fail with error message
can't read "dataset_name": no such variable
There is an updated ccp4i file (further updated on 23/03/2009) which should fix this problem:
(Thanks to Matsuura Takanori and Nick Bauer)
Running imosflm exits with
$ imosflm MOSDIR is Error in startup script: invalid command name "iwidgets::tabnotebook" while executing "iwidgets::tabnotebook $itk_interior.f.tabs -tabpos n -background "#dcdcdc" -tabbackground "#a9a9a9" -foreground "black" -tabforeground "black" ..."
This is due to tclIndex not being copied from tcltk++/iwidgets4.0.1/generic to the <install loc>/lib/iwidgets4.0.1/scripts during the build-tcl-tk++.sh process. The solution is to manually copy the file.
This problem occurs when running Pisa from the command line or through the CCP4i interface. Using the "Analyse Protein Interfaces" task under "Structure Analysis" will give rise to the error message "cannot load assembly analysis parameters". The error is due to path variables in the pisa.cfg file for identifying the location of CCP4 being hard-coded to "/usr/local/xtal/ccp4-6.1.0". To get around the problem the user should manually edit the file $CCP4/share/pisa/pisa.cfg to have the correct path names or they can run the script linked to below to perform the correction to the file. This requires the user to have write permissions in the CCP4 installation area. Note that on Mac OSX systems the user will need to use "sudo -s" to run the script. .
Pisa config file correction script:
(Thanks to Paul Emsley for finding this)
There is a minor formatting bug in the acorn.f source code file. This results in failure of the program but only in the unusual case when the user has selected to use one of the more obscure options in acorn. The link below is to a patch file which will fix the source code file.
Patch file for acorn.f:
(Thanks to Eleanor Dodson for finding this)
This problem occurs when trying to start iMosflm from the CCP4 interface without having set up a project and project directory in CCP4i. iMosflm needs a project directory set to output its files to. If you see this error go to the "Directories&ProjectDir" menu and ensure that you have at least one project specified. This problem usually only occurs after a new installation of the suite has been performed.
Clicking on the sftools task results in a Tcl error, and the window fails to open.
This is a side effect of a fix of different problem. There are 2 updated ccp4i files which should fix both problems:
Refmac ignores any input supplied using the "Include keyword file" option.
There is an updated ccp4i file which should fix this problem:
(Thanks to Dima Klenchin)
Compilation with a default Fortran optimisation level of -O2 causes programs which use Fast Fourier Transforms to fail, because the compiler does not perform integer division correctly. The fix is to reduce optimisation to -O1.
The fix is available as a patch configure-r1.372.2.18-r1.372.2.19.diff (also available from ftp://ftp.ccp4.ac.uk/ccp4/6.0.2/patches) to $CCP4/configure.
Note that compilation using Intel 9.0 compilers fails entirely, apparently because of missing symbols in the compiler's C library.
Compilation of Refmac5.2 with gfortran 4.1.1 (and possibly other versions?) fails with:
ccp4e.log:105:make: *** [make_unix_ccp4.o] Error 1 ccp4e.log:106:make: *** [srcdir] Error 2 ccp4e.log:115:make: *** [make_unix_ccp4.o] Error 1
The fix is to change IDATE into UIDATE because gfortran 4.1.1 IDATE specifications have changed, UIDATE is a ccp4 routine which does not change:
.../ccp4/src/refmac5_$ cvs diff -r1.10 -r1.11 make_unix_ccp4.f diff -r1.10 -r1.11 100c100 < CALL IDATE(IMON,IDAY,IYR) --- > CALL UIDATE(IMON,IDAY,IYR)
MTZ2VARIOUS used to output a header to the SHELX file. The idea was that this was a template for the required foo.ins file. However, this is not convenient and has been removed. Reflections in the free-R set were also flagged with "FREE" which can no longer be read by SHELX, so this has been removed. Output from the FSQUARED option was automatically scaled to avoid format overflow. This has now been applied to intensity output as well. Finally, it was possible to use "LABIN I=I SIGI=SIGI" which is better than the FSQUARED option, but this was erroneously flagged as HKLF 3 output. This is now fixed.
Updated files are available:
(Thanks to Ian Tickle)
There is an error in the section of the CCP4 installation documentation dealing with the automated installer script for UNIX and Linux platforms, e.g. at http://www.ccp4.ac.uk/dist/INSTALL.html#install_sh.
The documentation refers to a "configure" subdirectory that should be created as part of the installation process, and which should contain the setup files (e.g. ccp4.setup) for the installation.
However this information is no longer correct. The revised text should read:
The script will also generate a setup-scripts subdirectory that contains set-up files for the various packages (i.e. ccp4.setup for the core suite and programs, and ccp4-others.setup for CCP4mg and Coot) for Bourne/Bash shell (in the sh subdirectory of setup-scripts) and C-shell (in the csh subdirectory). To start using the software you should only need to source the appropriate set-up files for your shell.
The installation documentation has been updated for future releases of the suite.
(Thanks to Susumu Ichiyama)
The phaser 2.1.3 update includes a updated ccp4_general_f.c which calls ccp4_flush_stdout which should be in unix.m4. A replacement unix.m4 is available.
The output from mtz2various in scalepack format was incorrect for pairs where I- was measured but I+ was not. Also fixes lack of meaningful output when intensities are read rather than structure factor amplitudes. mtz2various.f-r126.96.36.199-r1.136.diff (also available from ftp://ftp.ccp4.ac.uk/ccp4/6.0.2/patches) to $CCP4/src/mtz2various.f and remaking mtz2various.
(Thanks to Pete Dunten)
Topp claims it cannot find the input files corresponding to keywords MOL1 and MOL2, even though the files exist.
This can be fixed by applying the patch topp.f-r188.8.131.52-r184.108.40.206.diff (also available from ftp://ftp.ccp4.ac.uk/ccp4/6.0.2/patches) to $CCP4/src/topp_/topp.f and remaking topp.
An extra space has been added to the output line for the analysis against resolution, Mn(I/sd) from Scala to allow for values of I/sigma equal to or greater than 100. This patch is particularly important for users of XIA.
This can be fixed by applying the patch scala.f-r220.127.116.11-r18.104.22.168.diff (also available from ftp://ftp.ccp4.ac.uk/ccp4/6.0.2/patches) to $CCP4/src/scala_/scala.f and remaking scala.
There is also an updated Scala windows executable available from here: scala.exe
(Thanks to Graeme Winter)
Polarrfn is sensitive to the ordering of its input HKLIN files. If the order is one way the program works fine but if the files are reversed the program exits prematurely with an end of file error.
This can be fixed by applying the patch polarrfn.f-r22.214.171.124-r126.96.36.199.diff (also available from ftp://ftp.ccp4.ac.uk/ccp4/6.0.2/patches) to $CCP4/src/polarrfn.f and remaking polarrfn.
(Thanks to Pete Dunten)
The number of fixed translation solutions allowed as input to Amore was limited to 8. Consequently one was restricted to finding no more than 9 molecules in the asu (8 fixed + 1 new). No error message was given if this limit was exceeded. Instead the output Euler angles and fractional translations became corrupted, due to array boundaries being exceeded.
The limit has been increased from 8 to 64, and an error message is printed if the new limit is exceeded. The translation function output has been changed slightly to allow for two digit labels. The changes are available as a patch: amore.f-r1.115-r1.116.diff.
Note added 26/11/07: If running through ccp4i, the following patch should also be installed: amore_utils.tcl-r1.14-r1.15.diff.
(Thanks to Adrian Lapthorn)
The version of cmakereference distributed with CCP4 6.0.1/2 requires compressed .ent.Z files. Sometimes an uncompressed file may be downloaded from an archive site. This should be compressed using compress before using cmakereference.
Note: the supplied reference data in $CLIBD/reference_structures should be suitable in all but the most exotic cases.
There are some minor format errors in the Isomorphism Correlation table output by bp3, which cause loggraph and xloggraph to fail when used to view this data.
These can be corrected by use of the patch crystal.C-r1.5-r1.6.diff.
(Thanks to Graeme Winter)
When running SORTMTZ for example as part of the "Scale and Merge Intensities" (SCALA) task, the program may fail with the message "Too many records".
This has been reported when the MTZ file provided as input is very large (the file causing the current problem contained of the order of 2.5 million reflections in 70 batches).
This has been reported on Windows systems, where there may be similar problems with CAD, SCALEIT and SCALA itself. The fix for Windows users is to obtain the compressed .zip file from the CCP4 FTP server:
This file contains new versions of cad.exe, sortmtz.exe scala.exe and scaleit.exe, which should be put into the "bin" directory within your CCP4 installation to replace the distributed versions.
(Thanks to Richard Strange, Graeme Winter and Francois Remacle)
The cif file created by mtz2various contains the "_audit.creation_date" item in "yyyy-dd-mm" format. It should actually be "yyyy-mm-dd" according, to the mmCIF dictionary:
The following patch to mtz2various.f corrects the writing of the date information:
(Thanks to Takaaki Fukami)
There is a small bug that crops up on some occasions with the 6.0.2 version of Refmac5. One of the eigenvector calculations encounters an error and causes the program to terminate with the error message "Problem in aniso_eigens #", where # is some integer value. The latest version of refmac available from the York YSBL site fixes this and is available from:http://www.ysbl.york.ac.uk/~garib/refmac/latest_refmac.html.
The option of inputting already squared values (e.g. F2OR/E2OR from ecalc) into FFT, by assigning it to I instead of F1, causes FFT to write out some incorrect information. It also causes it to not switch to the correct Patterson space-group. This can be corrected by applying the following patch to fft.f:fft.f-r1.98-r1.100.diff.
(Thanks to Clemens Vonrhein)
Running rstats from the autoSHARP script may result in divide by zero errors. This is because the autoSHARP script only outputs a subset of the input mtzfile columns into the output mtzfile which "corrupts" the internal representation of the mtzfile statistics. The fault lies entirely within rstats.
This can be corrected by applying the following patch to rstats.f:rstats.f-r1.39-r1.40.diff.
(Thanks to Markus Meier)
CAD may crash when an integer name is used as the crystal name (XNAME), otherwise, the behaviour is ill defined. It is strongly suggested that a pure integer not be used as the crystal name as this clashes with parsing within CAD, and possibly other CCP4 programs, with respect to the Dataset ID mechanism of accessing columns. A character should be contained within the XNAME.
There is a typo error in one of the MPI_BCast calls in Beast_MPI. A patch to fix this can be found at the link below. (Note that beast is now deprecated from the suite and will not be included in releases after CCP4 6.1).
(Thanks to Maier Timm)
Refmac only allows ten characters each for the Project and Dataset names used for harvesting. Names longer than this will be truncated, which can lead to error messages when running from the ccp4i gui.
The following patch increases the length allowed for these names to 64, in accordance with the MTZ file specification:
(Thanks to Ingo Korndoerfer and Herbert Klei)
There is a bug with running a CCP4i job using the "Run and edit com file" option when the input file name is of the form foo.bar.mtz or similar (for example rnase.1.mtz). The job starts running but generates a window with an error message of the form:
bad window path name ".edit_rnase"
The job stops but its status remains as RUNNING. This will happen for any file with multiple "dots" in the name, regardless of extension.
There are a couple of possible work-arounds:
Alternatively, apply the following patch from the CCP4 ftp server to the file $CCP4/ccp4i/src/util_windows.tcl:
To clean up a job that still appears to be running after this bug has occurred, select the job in question and then go to the Edit Job Data option under the menu of the same name (on the right-hand side of the main CCP4i window). You can then change the job status to FAILED.
(Thanks to Randy Read)
H-L coefficients can be specified on the LABIN keyword of mtz2various when converting to X-PLOR/CNS format. However, this functionality is not available in the GUI.
It is now. Fixed ccp4i files (also includes SHELX output fix):
(Thanks to Bernie Santarsiero, Vineet Gaur)
Attempting to start Detwin from ccp4i on windows gives an error message
Error: wrong #args: should be "CreateTaskWindow arrayname title icon_name ?frame_list? args"
This can be fixed by replacing the detwin.tcl in %CCP4%/ccp4i/tasks with a modified detwin.tcl.
If "PDB file output" or "MTZ file output" are selected in the last folder "Additional parameters" then Phaser will fail with "SYNTAX ERROR: Use ON or OFF". This is because ccp4i adds quoting to ON or OFF which Phaser can't interpret.
The quoting has been removed. A patch file is available at:
(Thanks to Serge Cohen and Randy Read for alerting us.)
Under the circumstances that there is a pre-existing ...com.tmp file in $CCP4_SCR area with the same name as a generated file, ccp4i will append the input data to the pre-existing file. This may cause unexpected results. An example appending file for a unique job is shown:
CELL 117.5058 125.8665 245.4758 90.0000 90.0000 90.0000 SYMMETRY 'P 21 21 21' LABOUT F=FUNI SIGF=SIGFUNI RESOLUTION 2.200 ## This script run with the command ########## # unique HKLOUT "/tmp/ccb/PROJECT1_25_3_mtz.tmp" ################################################ CELL 117.5058 125.8665 245.4758 90.0000 90.0000 90.0000 SYMMETRY 'I 21 21 21' LABOUT F=FUNI SIGF=SIGFUNI RESOLUTION 2.199 ## This script run with the command ########## # ################################################
The solution is to rerun the job and/or delete the pre-existing ...com.tmp files.
Running Refmac via the ccp4i gui and asking it to generate weighted difference maps will result in the warning: "no file name for parameter MAPOUT1".
There are some known issues when running CCP4i using the latest version of Tcl/Tk (version 8.5), in particular the appearence of check buttons or toggle buttons.
Installation of Phaser binary has been reported to fail on some systems running SuSE 10, the error log contains messages of the form:
ccp4e.log:7:mkdir: error while loading shared libraries: libc.so.6: cannot open shared object file: No such file or directory
This appears to be due to the environment variable LD_ASSUME_KERNEL being set inappropriately by the $CCP4/src/phaser/conf/configure-options file.
The fix is to apply the following patch to the file:
Note that the file needs to be updated before the installation is attempted, so it is recommended that you remove the failed installation files and re-unpack the suite before patching this file and installing again.
This bug should be fixed in subsequent versions of CCP4.
the PDB file output from anisoanl is empty apart from the END line.
This problem can be fixed by applying the following patch anisoanl.f-r1.22-r1.24.diff
You specify to exclude an atom using the at_excl keyword, but sc still fails if that atom is in the input coordinates and not in the SCRADII file.
This is because the at_excl keyword is not processed until the input coordinates have been read and their radii looked up in the SCRADII file. This problem can be fixed by applying the following patches. sc.f-r1.9-r188.8.131.52.diff and setup.fh-r1.5-r184.108.40.206.diff
Chainsaw fails when modifying a C-terminal residue which contains atoms named OT1 and OT2. Such a file can be generated by using Moleman2 to split a pdb file containing NMR data into individual models.
This can be fixed by applying the patch chainsaw.cpp-11May2006.diff (also available from ftp://ftp.ccp4.ac.uk/ccp4/6.0.1/patches) to $CCP4/src/mmdb_app_chainsaw.cpp and remaking chainsaw.
(Thanks to Nick Keep)
Contact may give junk output when the input pdb file contains more than about 32000 atoms.
This can be fixed by applying the patch contact.f-12June2006.diff (also available from ftp://ftp.ccp4.ac.uk/ccp4/6.0.1/patches) to $CCP4/src/contact.f and remaking contact.
(Thanks to Andrew Leslie)
Several problems have been encountered with the program ACT. The temperature factor analysis stops with error message
ACT: MOMENT: n must be at least 2if it encounters a chain containing only one residue.
In addition, most HETAMs with three letter names are erroneously taken to be amino acids, causing them to be included in the Main chain/Side chain statistics.
A number of array limit failures can also occur with pdb files containing a large number of atoms.
Fixes have been incorporated in the patch act.f-6July2006.diff (also available from ftp://ftp.ccp4.ac.uk/ccp4/6.0.1/patches).
(Thanks to Ketan Gajiwala)
A bug in rotmat means that it may sometimes incorrectly claim to have detected a hit and the rotation angles may be incorrect.
This can be fixed by applying the patch rotmat.f-17July2006.diff (also available from ftp://ftp.ccp4.ac.uk/ccp4/6.0.1/patches) to $CCP4/src/rotmat.f and remaking rotmat.
(Thanks to Jean-Paul Declercq for spotting this bug and Eleanor Dodson for fixing it.)
When molrep is run using an electron density map as the model, the map may be read in incorrectly, due to an uninitialised variable in the code. This sometimes gives rise to the following error message
ERROR: Rad_inertia of model = 0 (ROTDNS)
The problem can be fixed by applying the patch molrep.f-r1.28-r220.127.116.11.diff (also available from ftp://ftp.ccp4.ac.uk/ccp4/6.0.1/patches) to $CCP4/src/molrep_/molrep.f and remaking molrep.
Alternatively the latest version 9.4.05 of Molrep can be downloaded from York University.
(Thanks to Joseph Ho for reporting this problem.)
In all versions of CCP4 later than 4.2.2, CAD fails to copy the history lines from the input mtz file(s) to the output mtz file. OMIT and MTZTONA4 suffer similar problems.
Occasionally, memory faults can occur if there is a large number of history lines in the input.
Polarrfn can exhibit a failure in which it appears to run correctly but the map contains only NaN.
Weighted mean of map NaN Weighted rms 0.000
This can be fixed by the simple one line change to polarrfn:
149c149 < REAL ADATA(NLPRGI),COLRNG(2,NLPRGI) --- > REAL ADATA(NLPRGI),COLRNG(2,MCOLS)or through polarrfn.f-r1.45-r1.46.diff.
Some fixes for small bugs in Detwin have been provided by Pete Dunten. These include stopping Detwin from trying to detwin data when it does not have all of the reflection data required for detwinning and also a fix to prevent the java applet from breaking when no title is assigned and no twin fraction is assigned.
The updated source code can be found here: detwin.f.
(Thanks to Pete Dunten for reporting this problem and providing the updates to the source code.)
Running Refmac via the ccp4i gui and asking it to generate weighted difference maps will result in the warning: "no file name for parameter MAPOUT1".
This can be fixed by replacing the file refmac5.tcl in the ccp4i/tasks folder with this one: refmac5.tcl.
(Thanks to Kinlin Chao and Nick Keep for pointing this one out)
In some circumstances the CCP4 parser function ccp4_parse may identify comment characters in lines even though no such characters are present, and as a result the line is incorrectly tokenised. The circumstances in which this behaviour occurs are not well understood however it is likely that in most cases it is not a problem.
The fix is to initialise the iscommt flag at the head of the ccp4_parse function. The following patch contains this fix (in addition to new diagnostic codes) and should be applied to the $CLIBS/ccp4_parser.c file:
Compilation fails on Suse 8.2 system with gcc 3.3. Example error output is below, followed by a workaround.
> ------------------------------------------------------------ > **** Making regex-v2.1 **** > ------------------------------------------------------------ > make: Entering directory `/mntdirect/_scisoft/pxsoft/src/ccp4/ccp4-6.0.0/suse82/ccp4-6.0/src/harvest_app_/pdb_extract/regex-v2.1' > gcc -O -Werror -Wall -DHAVE_STRCASECMP -DINCL_TEMPLATE_SRC -DHAVE_PLACEMENT_NEW -DNO_RANGE_CHECK -I./include -I../include -DPOSIX_MISTAKE -c ./src/regcomp.c -o obj/regcomp.o > src/regcomp.c: In function `freeset': > src/regcomp.c:1061: warning: comparison between signed and unsigned > src/regcomp.c: In function `freezeset': > src/regcomp.c:1092: warning: comparison between signed and unsigned > src/regcomp.c:1095: warning: comparison between signed and unsigned > src/regcomp.c: In function `firstch': > src/regcomp.c:1119: warning: comparison between signed and unsigned > src/regcomp.c: In function `nch': > src/regcomp.c:1139: warning: comparison between signed and unsigned > make: *** [obj/regcomp.o] Error 1 > make: Leaving directory `/mntdirect/_scisoft/pxsoft/src/ccp4/ccp4-6.0.0/suse82/ccp4-6.0/src/harvest_app_/pdb_extract/regex-v2.1' > make: *** [compile] Error 1 > make: Leaving directory `/mntdirect/_scisoft/pxsoft/src/ccp4/ccp4-6.0.0/suse82/ccp4-6.0/src/harvest_app_/pdb_extract' > make: *** [pdb_extract_suite] Error 2 > make: Leaving directory `/mntdirect/_scisoft/pxsoft/src/ccp4/ccp4-6.0.0/suse82/ccp4-6.0/src' > make: *** [srcdir] Error 2 > tarzan1:/scisoft/pxsoft/src/ccp4/ccp4-6.0.0/suse82/ccp4-6.0
Workaround: You will need to configure and compile manually. After unarchiving the source distribution, remove the "-Werror" from [ccp4_directory]/src/harvest_app_/pdb_extract/etc/make.platform.gnu3. Then configure and compile as usual, but DO NOT USE the --non-shared option.
The CCP4 Python binaries provided by CCP4 don't work with Suse 10. Running cctbx.python gives a libreadline import error.
Python 2.4.2 (#2, Jan 26 2006, 11:09:53) [GCC 3.2 20020903 (Red Hat Linux 8.0 3.2-7)] on linux2 Type "help", "copyright", "credits" or "license" for more information. Traceback (most recent call last): File "/etc/pythonstart", line 7, in ? import readline ImportError: libreadline.so.4: cannot open shared object file: No such file or directory
Suse 10 includes libreadline.so.5, but no libreadline.so.4. Create a link by doing the following (as the root user).
ln -s /lib/libreadline.so.5 /lib/libreadline.so.4
There is a problem which causes pirate to fail when applying ncs in any case where there is a proper two fold axis.
This can be fixed by applying the patch pirancslib.cpp-23Feb2006.diff (also available from ftp://ftp.ccp4.ac.uk/ccp4/6.0.0/patches) to $CCP4/lib/clipper/src/pirancslib.cpp and reinstalling cpirate.
(Thanks to Kevin)
Geomcalc fails with an RWBROOK error when running the test example in $CEXAM.
This can be fixed by applying the patch geomcalc.f-28Feb2006.diff and remaking geomcalc.
The truncation of the input weights to integers in cphasecombine was done incorrectly.
This can be fixed by applying the patch cphasecombine.cpp-09Mar2006.diff and reinstalling cphasecombine.
(Thanks to Kevin)
The findncs example in %CEXAM% (equivalent to $CEXAM/unix/runnable on unix based systems) gets stuck in an infinite loop in the refinement subroutine refrt in lgglib.f.
This has been fixed by terminating refinement once the rms value stops decreasing significantly or a preset number of refinement cycles is exceeded. The same problem was also seen on Linux systems with Intel compilers, where it could be avoided by reducing the optimisation level to -O0.
The xplot84driver gui is corrupted/does not work on Suse 8.2 and Redhat Enterprise 3.
To avoid this issue on Suse 8.2 install from source not from binaries. See the problem about compiling ccp4 6 on Suse 8.2 for how to do this.
When clicking on the "Shelx C/D/E HA Search/Phasing" button, the task doesn't start but instead gives an "Error in Tcl Script" pop-up window. Dismissing the pop-up and then trying again results in a partial window for the Shelx task.
This appears to be due to an incompatibility between Tcl 8.3 and Tcl 8.4, and should only be observed by people using the former version. One fix is to apply the following patch to the $CCP4/ccp4i/src/CCP4_utils.tcl file:
Alternatively you can upgrade the version of Tcl/Tk that you are using to run CCP4i with, to version 8.4 - see:
(Thanks to James Vergis)
Selecting 'Apply no editing to model' in the molrep gui does not have the desired effect of setting 'SURF N' in the com file. Instead the program defaults to 'SURF Y'.
This can be fixed by applying the patch molrep.com-13Mar2006.diff and restarting CCP4i.
(Thanks to David Borhani, Charlie Bond)
I was trying to use RSPS from the CCP4i GUI. When I run 'find sets of sites with good cross vectors' all is going well. But when I want to use fixed sites for the vector search 'find sites with good cross vectors to fixed site(s)' I am asked to supply a name for the parameter XPEAKSFIL. Unfortunately, there is no input field for such a filename in this RSPS mode.
Fix to $CCP4/ccp4i/tasks/rsps.tcl Associated change in $CCP4/ccp4i/scripts/rsps.script These will be in release 6.0.1
(Thanks to Albrecht Messerschmidt for reporting what seems to be a long-standing problem)
Various ccp4i tasks which rely on the parsing of log files may fail due to the lack of synchronisation between the fortran and C/C++ outputs, eg amore.
The workaround is to set GFORTRAN_UNBUFFERED_ALL to either 1 or Y at runtime, to turn off buffering of output (there will be some performance penalty for this).
(Thanks to Marian Szebenyi for reporting this)
When I try to (re)run a Refmac job, the task window does not open, and I get an error dialog box saying:
Error: can't read "array(RUN_COOT_FW)": no such element in array
RUN_COOT_FW is a new variable in 6.0. This can go wrong if the user has his own version of the .def file in $HOME/.CCP4/CCP4I_TOP/tasks/refmac5.def. The solution is to remove this file.
Linking against libccp4f.so may fail for some linuces when using gcc4 with gfortran
gfortran -O -w -fno-second-underscore -fPIC -o abs /data/sware/progs/ccp4- 6.0/src/abs.f -L../lib/src -lccp4f -lccp4c -lmmdb -L/data/sware/progs/ccp4- 6.0/lib/ccif -lccif -L/data/sware/progs/ccp4-6.0/lib/lapack -llapack - L/data/sware/progs/ccp4-6.0/lib/lapack -lblas -lstdc++ -Wl,- rpath,/data/sware/progs/ccp4-6.0/lib -lm ../lib/src/libccp4f.so: undefined reference to `getpid_' collect2: ld returned 1 exit status make: *** [abs] Error 1
There is a known workaround and a fix. The workaround is to use configure linux instead of configure --with-shared-lib linux. The fix is to apply the patch ccplib.f-26Feb2006.diff (also available from ftp://ftp.ccp4.ac.uk/ccp4/6.0.0/patches) to $CCP4/lib/src/ccplib.f and to remake the libraries.
A bug fix present in Clipper 1.5 and 2.0, was inadvertently not included in the 1.0 version supplied with CCP4.
The fix is to apply the patch derivs.h-06Mar2006.diff and remake clipper.
(Thanks to Tim Gruene)
Phaser via ccp4i apparently fails with FILE OPENING ERROR. The header will report:
*** Phaser Module: READ DATA FROM MTZ FILE 2.0 ***This is in fact from Phaser 2.0 distributed with Phenix. CCP4 6.0.2 has Phaser version 1.3.3, while CCP4 6.1 will likely have Phaser 2.1
Solution: if you want to use the CCP4 version, make sure you have source'd the ccp4.setup last, and that the CCP4 bin area is first on your path.
If using the CCP4 SuSe 9-9.2 executables for Phaser/CCTBX with the native SuSe 10 Python, when using CCTBX in Python you may get an error which ends like this:
ImportError: .../ccp4/6.0/ccp4-6.0/src/phaser/phaser-1.3.2/build/intel-linux/lib/libboost_python.so: undefined symbol: PyUnicodeUCS2_AsWideChar
You may get a similar error on some other systems. If so, please let us know and look at the possible solutions below. This is because the unicode encoding in your native Python is not compatible with that which was specified in compiling the CCTBX Python binding.
There is a small problem with compiling cctbx with Python version 2.5 (also 2.5.1). It results in the following error from the compiler in relation to the file objects_protocol.cpp in the folder cctbx/cctbx_sources/boost/libs/python/src
error: cannot convert int* to Py_ssize_t* for argument 2 to int _PyEval_SliceIndex(PyObject*, Py_ssize_t*)A patch for this file has been created to fix this problem and can be downloaded from here: object_protocol.cpp-r1.1-r1.2.diff.
(Thanks to Chris Fromme)