 |
Collaborative Computational Project No. 4 |
 |
|---|
|
|
Collaborative Computational Project No. 4 |
|
|---|
| Home | About CCP4 | CCP4 Projects | Downloads | Documentation | Courses | Developers | CCP4 people | WG1/WG2 | Privacy |
|---|
These are the known problems with release 4.2 of the CCP4 suite, together with fixes where appropriate. Code changes are given as patches, which can be applied with the UNIX patch utility. New files are put in the prerelease directory, which can be found at ftp://ftp.ccp4.ac.uk/prerelease.
Hints and advice about applying the fixes can be found here.
If you can't find your problem here, please email us.
Warning for Linux users: if you are running CCP4 compiled with gcc 2.96 may be unreliable.. This has been seen with RedHat 7.0 and Mandrake 8.1.
A list of general and version-independent problems with CCP4i can be found here:
When ccp4 is build using --with-shared-lib the mosflm build can fail with
Archive: ../jpg/libjpeg.a has no table of contents (not searched)this is due to the forced libjpeg.a make in the Mosflm makefile. A work around, after failed build, is as follows:
cd $CCP4/x-windows/Mosflm/jpg make cp .libs/libjpeg.a . ranlib libjpeg.a cd .. make cd $CCP4 make install
A typo in molrep.f leads to a scalar variable masquerading as an array.
This may lead to the program failing with a sigsev error (though other failures
have also been reported, for example: on Mac-OSX an 'Abort trap' may occur whilst
reading sequential formatted internal IO).
The fix is to apply the following patch and recompile molrep.
diff -c -3 -r1.12.2.1 molrep.f
*** molrep.f 2002/10/07 14:17:53 1.12.2.1
--- molrep.f 2002/11/27 11:29:53
***************
*** 9303,9309 ****
ENDDO
ENDIF
! CALL GET_CR_MATRIX(FRTOCR,CRTOFRF)
IF(PROG.EQ.'CROS'.OR.PROG.EQ.'CRSL') THEN
--- 9303,9309 ----
ENDDO
ENDIF
! CALL GET_CR_MATRIX(FRTOCR,CRTOFR)
IF(PROG.EQ.'CROS'.OR.PROG.EQ.'CRSL') THEN
(Thanks to Phil Evans)
change to scala.f 17663c17663 < $ nrfrej, irogue, nstrej(13), nrunp1, ibtanl, ip, irn, nmpr, --- > $ nrfrej, irogue, nstrej(25), nrunp1, ibtanl, ip, irn, nmpr, 18172c18172 < do 18, i = 1,13 --- > do 18, i = 1,25(Thanks to Phil Evans and Frank Vondelft)
When Refmac 5.1.24 (or earlier) is run with a high resolution cutoff > 1.0A, a bug in the program forces it into an infinite loop - it appears to be running, but produces no output after reading in the libraries and will never finish.
The workaround is to avoid using a high resolution cutoff above 1.0A. This bug will be fixed in future releases of Refmac5.
(Thanks to Sue Roberts and Garib Murshudov)The Molrep documentation states that you must use the PHIC label to specify phases, however this is incorrect - the program actually expects PHIB.
(Thanks to Stephen Soisson)(Thanks to Pete Dunten)
diff -c -r1.63 scala.f
*** scala.f 25 Jun 2002 16:24:11 -0000 1.63
--- scala.f 2 Jun 2003 07:53:40 -0000
***************
*** 10214,10219 ****
--- 10214,10220 ----
if (lflag .eq. 0) then
c A number read
n = n+1
+ lcorrc = .true.
endif
elseif (cvalue(n) .eq. 'CHEC') then
kprflo = -min(iabs(kprflo), 10)
***************
*** 10624,10629 ****
--- 10625,10631 ----
c USE_PROFILE use profile-fitted intensity even for scaled partials
c
410 n = 2
+ lfinal = .true.
lcorrc = .false.
lprtst = .false.
c lprscl is flag that SCALE_PARTIAL has been set
***************
*** 10683,10688 ****
--- 10685,10691 ----
prthou = prthld
n = n+1
endif
+ lcorrc = .true.
elseif (cvalue(n) .eq. 'USE_') then
c Use profile-fitted intensities for scaled partials (only if present)
if (ncip .gt. 0 .and. ncsdip .gt. 0) then
Description TRUNCATE may fail when running with data harvesting turned on, for example with the error message child killed: bus error.
Fix/Workaround Currently there is no fix for this problem. The program will run if data harvesting is turned off.>
CCP4i may fail to start on some Windows configurations, complaining that the HOME environment variable is not set or that it doesn't point to a valid directory. In fact, CCP4i doesn't need the HOME variable so this check is unnecessary.
The fix is to download updated versions of $CCP4/bin/ccp4i.tcl and $CCP4/src/system.tcl file from the prerelease area, where they can be found as ccp4i.tcl-4.2.2 and system.tcl-4.2.2.
(Thanks to Jennifer Durkee for bringing this to our attention.)The option to reset HKL limits in the output MTZ file (in the ``Define MTZ Output folder) is incorrectly implemented. The correct implementation is for eitherthe Laue code or explicit HKL limits to be defined.
The fix is to download updated versions of $CCP4/ccp4i/tasks/cad.tcl, $CCP4/tasks/cad.def and $CCP4/templates/cad.com files from the prerelease area, where they can be found as cad.tcl-outlim, cad.def-outlim and cad.com-outlim.
(Thanks to Eleanor Dodson for identifying the problem.)CCP4i may have difficulty opening a task window, giving a stack trace
can't read array(_element): no such element in array while executing "catch "set type $array($_element)" ...This is due to a difference in the effect of quotes versus braces between tcl8.4 and 8.3. The solution is to replace catch "..." with catch {...}. For the failure listed above the following changes should be made to ccp4i/src/utils.tcl
*** utils.tcl 2002/04/04 14:35:59 1.11
--- utils.tcl 2003/01/21 13:26:35 1.12
***************
*** 293,301 ****
set typelist {}
if { [llength [set l [split $element ,] ] ] > 1 } {
! set status [catch "set type $array(_[lindex $l 0],0)"]
} else {
! set status [catch "set type $array(_$element)"]
}
if { !$status && $type != "" && [info exists typedef($type)] } {
--- 293,301 ----
set typelist {}
if { [llength [set l [split $element ,] ] ] > 1 } {
! set status [catch {set type $array(_[lindex $l 0],0)}]
} else {
! set status [catch {set type $array(_$element)}]
}
if { !$status && $type != "" && [info exists typedef($type)] } {
The specification of output MTZ labels are reversed for the two options of combining experimental and partial models and combining MIR data.
The following changes should be made to $CCP4/ccp4i/tasks/sigmaa.tcl:
*** sigmaa.tcl 2001/02/28 10:02:37 1.4
--- sigmaa.tcl 2003/02/03 09:48:32 1.5
***************
*** 284,290 ****
CloseSubFrame
! OpenSubFrame frame -toggle_display SIGMAA_ACTION open COMB_PART
CreateLaboutLine line \
"Combined phase angle and weight" \
--- 284,290 ----
CloseSubFrame
! OpenSubFrame frame -toggle_display SIGMAA_ACTION open COMB_MIR
CreateLaboutLine line \
"Combined phase angle and weight" \
***************
*** 306,312 ****
CloseSubFrame
CloseSubFrame
! OpenSubFrame frame -toggle_display SIGMAA_ACTION open COMB_MIR
CreateLaboutLine line \
"Combined Phase and weight" \
--- 306,312 ----
CloseSubFrame
CloseSubFrame
! OpenSubFrame frame -toggle_display SIGMAA_ACTION open COMB_PART
CreateLaboutLine line \
"Combined Phase and weight" \
(Thanks to Hiroto Yamaguchi)
In Molrep, the keyword for inputting self-rotation function is FILE_TSR, but ccp4i writes FILE_SRF instead. The simple fix follows by change _SRF to _TSR:
69c69 < 1 _FILE_SRF $SROT_SOLUTIONS --- > 1 _FILE_TSR $SROT_SOLUTIONS(thanks to Jade)
The comments associated with the TEMPerature factor restraints are incorrect, if followed the behaviour will not be as expected
diff -r1.22 refmac5.tcl 1437c1437 < message "Enter weight for bonded Bfactor restraint (SIGB1)" \ --- > message "Enter weight for main chain bonded Bfactor restraint (SIGB1)" \ 1439c1439 < message "Enter weight for non-bonded Bfactor restraint (SIGB2)" \ --- > message "Enter weight for main chain non-bonded Bfactor restraint (SIGB2)" \ 1441c1441 < message "Enter weight for planar Bfactor restraint (SIGB3)" \ --- > message "Enter weight for side chain bonded Bfactor restraint (SIGB3)" \ 1443c1443 < message "Enter weight for Bfactor restraint involving an hydrogen atom (SIGB4)" \ --- > message "Enter weight for side chain non-bonded Bfactor restraint (SIGB4)" \Similarly the refmac5 documentation is incorrect; with the SIGB2 and SIGB3 definitions reversed.
The parser incorrectly treated comment lines and lines containing a single token before a comment. This lead to attempts to read from stdin when a non-blank input line was sent to the parser routine.
Apply the following patch to $CCP4/lib/src/parser.f
***************
*** 294,300 ****
NINCHR = 0
ENDIF
FIRST = .TRUE.
! GO TO 20
END IF
C
C---- Test if first field begins '@' and include file if so
--- 294,300 ----
NINCHR = 0
ENDIF
FIRST = .TRUE.
! IF(LENLIN.EQ.0) GO TO 20
END IF
C
C---- Test if first field begins '@' and include file if so
***************
*** 533,539 ****
+ (LINE(I:I).EQ.ICOMM1 .OR. LINE(I:I).EQ.ICOMM2)) THEN
COMMNT = .TRUE.
C special case; comment line:
! IF (N.EQ.1) THEN
N = 0
RETURN
END IF
--- 533,539 ----
+ (LINE(I:I).EQ.ICOMM1 .OR. LINE(I:I).EQ.ICOMM2)) THEN
COMMNT = .TRUE.
C special case; comment line:
! IF (I.EQ.1) THEN
N = 0
RETURN
END IF
The second reference given in the documentation has the incorrect page number. The reference should be:
Mosflm fails to link correctly against libccp4.so on linux. The resulting errors are of the form
/xtal/ccp4-4.2/lib/src/libccp4.so: undefined reference to `ccif_new_cif_'This would seem to be due to the linker loading the whole library, not just those modules necessary. The solution is to add -L<libccif location> -lccif to the MOSLIBS entry of mosflm/Makefile. A partial fix is in version 4.2.1. The latest Mosflm/configure and Mosflm/mosflm/Makefile.in are available mosflm_link.tar.gz. (Thanks to Todd Geders for informing us of the problem)
There is a potential problem with the LAPACK configuration when upgrading from 4.2 to 4.2.1.
If the original installation created LAPACK and/or BLAS libraries from the Netlib source code then it may be that configure detects these when upgrading, but fails to write the information to the Makefile headers - resulting in compilation errors for e.g. SCALA complaining that slamch or ssyev are unresolvable symbols.
The fix is to download the updated $CCP4/configure file from the prerelease area.
(Thanks to Sandra McLaughlin for informing us of the problem)When processing a mtz file containing batches belonging to a single dataset with
dataset number is 0 scala may experience runtime sigsev or bus errors. Behaviour on
other systems is undefined.
The fix is to apply the following patch to $CPROG/scala_/scala.f:
diff -c -r1.63 scala.f
*** scala.f 2002/06/25 16:24:11 1.63
--- scala.f 2002/07/12 08:41:35
***************
*** 22167,22174 ****
$ ' for batch ', ibatch
call ccperr(1,'*** Invalid batch dataset number ***')
else
! c Dataset ID unset
! j = 0
endif
c
5 jdtsbt(i) = j
--- 22167,22183 ----
$ ' for batch ', ibatch
call ccperr(1,'*** Invalid batch dataset number ***')
else
! c Dataset ID unset, set it to 1 (ie everything is one dataset)
! if (ndatasets .gt. 1) then
! c Can't do this if there is more than one dataset
! write (6, '(/a,a/)')
! $ ' ERROR: the file contains multiple datasets',
! $ ' but batches are not assigned to datasets (SETCRY)'
! call ccperr
! $ (1, '*** batches not assigned to datasets ***')
! endif
! j = 1
! idtsbt(i) = j
endif
c
5 jdtsbt(i) = j
This problem is also fixed in scala 3.1.11 from the
prerelease area as scala_3.1.11.tar.gz.
TIDY keyword broken.
The problem appears as below:Data line--- tidy orth 30 20 10 *** Warning No argument expected Data line--- END professs: Input error (see above) professs: Input error (see above) Times: User: 0.1s System: 0.1s Elapsed: 0:00Solution - make the following change:
101c101
< call parsekey('TIDY',tidyin)
---
> call parsesubkey('TIDY',' ',tidyin)
(Thanks to Pete Dunten for pointing this out)
LABOUT keyword broken.
You can omit the LABOUT keyword and change the labels later with CAD, or there is an updated program cif2mtz.f on the prerelease area.
COMBAT doesn't process XDET, YDET and ROT values correctly. This was originally reported for input from SAINT but is probably a bug for all input formats.
The fix is the following patch to be applied to $CPROG/combat_/combat.f:
*** combat.f 2002/06/14 14:31:30 1.14
--- combat.f 2002/08/27 10:04:21
***************
*** 939,951 ****
xdetmn = min(xdetmn, xdet)
ydetmn = min(ydetmn, ydet)
phimax = max(phimax, fphi)
- else
- C set ROT equal to batch.
- adata(13) = adata(5)
phimin = min(phimin, fphi)
adata(11) = xdet
adata(12) = ydet
adata(13) = fphi
endif
c write out reflection for monitoring
--- 939,951 ----
xdetmn = min(xdetmn, xdet)
ydetmn = min(ydetmn, ydet)
phimax = max(phimax, fphi)
phimin = min(phimin, fphi)
adata(11) = xdet
adata(12) = ydet
adata(13) = fphi
+ else
+ C set ROT equal to batch.
+ adata(13) = adata(5)
endif
c write out reflection for monitoring
(Many thanks to Toshihiko Akiba)
There is a minor bug when converting CNS Rfree to MTZ, if the file contains reflections with zero sigma. The program prints a warning and then fails to perform the conversion.
The bug is fixed by installing the updated file $CPROG/sftools_/sftools.f from the prerelease area.
(Thanks to Ezra Peisach and Bart Hazes)The ipdisp scritp has been found to fail with the detault Bourne shell on Tru64. This is due to a problem with the expansion of "$@". The solution is to replace "$@" with ${1+"$@"} for all occurences.
(thanks to Dave Love for finding the problem and providing the fix)
When using damping with PARTIAL, the damp keywork is not recognised if it is in lower case.
Workaround: use upper case.
Fix: apply the following patch to $CCP4/src/sigmaa.f.
diff -c -r1.51 -r1.50
*** sigmaa.f 2002/09/27 09:50:21 1.51
--- sigmaa.f 2002/08/08 13:20:22 1.50
***************
*** 556,563 ****
ICOMB = 0
ICHANG = 0
C DAMP D1 ( D2 ( D3) )
! N = 2
! CALL CCPUPC(CVALUE(N))
IF (CVALUE(N) .EQ. 'DAMP') THEN
ICHANG = 1
N = N+1
--- 556,562 ----
ICOMB = 0
ICHANG = 0
C DAMP D1 ( D2 ( D3) )
! N = 2
IF (CVALUE(N) .EQ. 'DAMP') THEN
ICHANG = 1
N = N+1
(Thanks to Yamaguchi Hiroto for providing the fix)
The OUTPUT PDB option of PEAKMAX doesn't align the atom name in the correct column of the output file. This causes unecessary warnings in other programs, for example WATPEAK, which use the file as input.
The following patch applied to $CPROG/peakmax.f fixes the problem:
diff -u -r1.41 peakmax.f
--- peakmax.f 2002/08/06 12:20:24 1.41
+++ peakmax.f 2002/10/02 14:34:01
@@ -216,7 +216,7 @@
+ NYc,NZEROPK,NZ1,NZ2,NZC,SX1,SX2,SY1,SY2,LFLAG,IST
LOGICAL LEND,LIMS,LRMS, LBFAC, LFIRST, LFRAC, LPEAK, XMLOUT
CHARACTER IDCH*1,ID*4,ATNAM*4,KEY*4,RESTYP*4,LINE*400,RESNO*4,
- + ALTCOD*1,SEGID*4,INSCOD*1
+ + ALTCOD*1,SEGID*4,INSCOD*1,ATMTYP*4
C ..
C .. Local Arrays ..
INTEGER MAXPK
@@ -309,6 +309,7 @@
LFRAC = .FALSE.
LPEAK = .FALSE.
ATNAM = ' O'
+ ATMTYP = ' O '
RESTYP = 'HOH'
IDCH = 'X'
BS(1) = 30.0
@@ -922,7 +923,7 @@
END IF
END IF
CALL XYZATOM(IXYZOUT,ISER,ATNAM,RESTYP,IDCH,IRES,RESNO,
- $ INSCOD,ALTCOD,SEGID,IZ,ID)
+ $ INSCOD,ALTCOD,SEGID,IZ,ATMTYP)
CALL XYZCOORD(IXYZOUT,'O','U',XX,YY,ZZ,QOCC,BISO,BS)
CALL XYZADVANCE(IXYZOUT,0,0,*310,*310)
IRES = IRES + 1
(Thanks to Phil Evans)
The keyword RAD has no effect, it is overridden by the default value. The fix follows, in $CCP4/src/molrep_/main_molrep_mtz.f:
1988c1988 < RAD = FVALUE(2) --- > ARAD = FVALUE(2)(Thanks to Yamaguchi Hiroto)
Contact will fail to identify contacts to main chain atoms in different chains if the residue number differs by less than 2.
The work around is to renumber the residues so that the numbers do not overlap. (Thanks to Mats Okvist
When CCP4i generates a keyworded script for FFT which excludes structure amplitudes below a certain minimum, it writes EXCLUDE F1MIN. In fact the correct keyword is EXCLUDE F1LIM.
Apply the following patch to $CCP4/ccp4i/templates/fft.com:
diff -r1.5 -r1.5.2.1
27,28c27,28
< -- {[IfSet $EXCLUDE_MINIMUM_1]} f1min $EXCLUDE_MINIMUM_1
< -- {[IfSet $EXCLUDE_MINIMUM_2]} f2min $EXCLUDE_MINIMUM_2
---
> -- {[IfSet $EXCLUDE_MINIMUM_1]} f1lim $EXCLUDE_MINIMUM_1
> -- {[IfSet $EXCLUDE_MINIMUM_2]} f2lim $EXCLUDE_MINIMUM_2
(Thanks to Eleanor Dodson)
There are two minor bugs:
The problems are fixed by updated versions of $CCP4/ccp4i/src/database.tcl and $CCP4/ccp4i/src/taskwindow.tcl, available from the prerelease area as database-lock.tcl and taskwindow-lock.tcl.
For the Prepare Data for HA task the wrong column labels are returned for SAD data as Dano outputing SHELX format. The correction is to apply the following patch to $CCP4/ccp4i/scripts/prephadata.script
diff -c -r1.5 prephadata.script
*** prephadata.script 2002/04/03 09:07:14 1.5
--- prephadata.script 2002/09/02 12:34:06
***************
*** 106,112 ****
set FP FM
set SIGFP SIGFM
} elseif { [StringSame $REVISE_INPUT SADDS] } {
! set LABIN "FPH SIGFPH FP SIGFP"
} elseif { [StringSame $REVISE_INPUT SIR] } {
set LABIN "FPH SIGFPH FP SIGFP"
} elseif { [StringSame $REVISE_INPUT NATIVE] } {
--- 106,116 ----
set FP FM
set SIGFP SIGFM
} elseif { [StringSame $REVISE_INPUT SADDS] } {
! set LABIN "DP SIGDP"
! unset DP
! unset SIGDP
! set DP $ECALC_DPH0
! set SIGDP $ECALC_SIGDPH0
} elseif { [StringSame $REVISE_INPUT SIR] } {
set LABIN "FPH SIGFPH FP SIGFP"
} elseif { [StringSame $REVISE_INPUT NATIVE] } {
Attempts to use SFALL generated structure factors with the
"combine isomorphous phase with partial solution" or
"combine two sets of MIR phases" fail as there are missing labels at the SIGMAA
stage. This is because only the F SIGF RFREE FC and PHIC labels are included in the
temporary mtz file.
The solutions are:
diff -r1.2 sigmaa.script
15a16,20
>
> if { $SIGMAA_ACTION != "PART" } {
> set ALLIN 1
> }
>
Two problems:
Both problems are addressed by applying the following patch to $CCP4/ccp4i/sketch/dictionary.tcl:
--- dictionary.tcl 2002/05/21 07:42:30 1.6
+++ dictionary.tcl 2002/09/10 14:37:40 1.6.2.2
@@ -426,7 +426,7 @@
for { set n 1 } { $n <= $Mol(nAtoms) } { incr n } {
append text " $Mol(chem_comp_id) [write_cif_name $Mol(Name,$n)]"
foreach output $output_list {
- append text " " $Mol($output,$n)
+ append text " " [string toupper $Mol($output,$n)]
}
append text \n
}
@@ -441,7 +441,7 @@
for { set n 1 } { $n <= $Mol(nBonds)} { incr n } {
set a1 [lindex $Mol(Bonds,$n) 0]
set a2 [lindex $Mol(Bonds,$n) 1]
- set type [lindex [list nowt single double triple] $Mol(Bondtype,$n)]
+ set type [lindex [list nowt single double triple delocalised aromatic metal coval] $Mol(Bondtype,$n)]
append text " $Mol(chem_comp_id) [write_cif_name $Mol(Name,$a1)] [write_cif_name $Mol(Name,$a2)] $type\n"
}
(Thanks to Liz Potterton)
There is an occassional problem when starting a job within CCP4i and then closing down the main interface while it runs - upon restarting the interface and going to the ``View Files from Job'' menu, some of the output files may not be listed.
This is particularly a problem for interfaces such as ARP/wARP, which rely heavily on registering output files with CCP4i whilst the task is running, however it could potentially occur with other tasks too. The log file for the job may also contain messages of the form:
ERROR running script can not connect to server port SERVER_HOST localhost SERVER_PORT 4441
A workaround is not to close the main CCP4i window until the job has finished running. A more useful fix can be effected by downloading updated versions of the files database.tcl, execute.tcl, task.tcl and taskbrowser.tcl (all should be put into the $CCP4/ccp4i/src/ directory). The files can be found on the prerelease area as database-buffer.tcl, execute-buffer.tcl, task-buffer.tcl and taskbrowser-buffer.tcl.
The ccp4i dmmulti interface will fail to run multiple domain jobs with the error message:
Error:can't read "array(MASKIN,1MASKIN,2)": no such element in array
The fix is to apply the following to $CCP4/ccp4i/tasks/dmmulti.tcl:
76c76 < append array(INPUT_FILES) "MASKIN,$n" --- > append array(INPUT_FILES) "MASKIN,$n "(Thanks to Yamaguchi Hiroto for the fix)
When inputing fixed monomers in the Molrep interface, the option fails (although the program runs to completion) and the log file contains the warning Unrecognized keyword: _MODEL_2
The error is because the file with the fixed monomers should be supplied on the command line rather than via a keyword. The following patches will fix the problem:
To $CCP4/ccp4i/scripts/molrep.script:
diff -u -r1.5 molrep.script
--- molrep.script 2002/04/23 09:32:36 1.5
+++ molrep.script 2002/09/30 13:48:29
@@ -59,6 +59,9 @@
if { [StringSame $MOLREP_MODE FIT] } {
append cmd " MODEL2 $XYZIN2"
}
+ if { [StringSame $IFFIXED 1] } {
+ append cmd " MODEL2 $XYZFIXED"
+ }
CreateComScript molrep molrep_script
set status [Execute $cmd $molrep_script program_status report]
To $CCP4/ccp4i/templates/molrep.com:
diff -u -r1.8 molrep.com
--- molrep.com 2002/04/23 09:32:39 1.8
+++ molrep.com 2002/09/30 13:49:14
@@ -17,9 +17,6 @@
ENDIF
ENDIF
-$IFFIXED _MODEL_2 $XYZFIXED
-{[IfSet $FUNCTION] && ![StringSame $FUNCTION A]} FUN $FUNCTION
-
IF { ![StringSame N $PSEUDO_TRANS_MODE] }
1 PST $PSEUDO_TRANS_MODE
{[IfSet $PSEUDO_TRANS] } VPST $PSEUDO_TRANS
(Thanks to Akihiro Okamoto for finding and fixing the problem.)
In certain circumstances Tcl errors of the form:
array(WINDOW) not foundmay appear when closing task windows. In other circumstances closing one window may cause a second window to close simultaneously. These problems have been observed when a task window has been invoked from the task list and at the same time another window has been opened to rerun the same task.
Both problems can be fixed by installing an updated taskwindow.tcl file in $CCP4/ccp4i/src/. The file can be found on the prerelease area as taskwindow-close.tcl.
NB. this updated file includes the fix from the previously reported problem regarding Corrupted database/database.LOCK.
Under certain circumstances there may be problems with recovering the correct parameters when rerunning an earlier BEAST job, for example: if the original job involved running a rotation search on molecule "mol2" then the rerun parameters will change this to "mol1".
The problem is fixed by installing an updated beast.tcl file in $CCP4/ccp4i/tasks/. The file can be found on the prerelease area as beast-rerun.tcl.
(Thanks to Randy Read for reporting the problem.)An uninitialized variable in LRASSN may cause program exit
Error in label assignmentswhen no user LABIN line is required (SCALA, COMBAT).
The fix is to apply the following to mtzlib.f:
***************
*** 545,550 ****
--- 545,551 ----
C---- First loop over user input labels and assign them to file labels
C If NLUSRI(MINDX)=0 then this loop is never executed
C
+ IERR = 0
DO 70 JDO50 = 1,NLPRGI
CWORK = LSUSRI(MINDX,JDO50)
IF ((NLUSRI(MINDX).GT.0) .AND. (CWORK.NE.' ')) THEN
This file is supposed to be used as example input to the program TRUNCATE. However, the file contains unmerged data (with batch information) and is not appropriate. In fact, the tutorial can be run with this file, but don't look too closely at the output!
If you are running configure irix ... on IRIX 6.5 then you will always get a shared version of libccp4 even if shared libs were not requested via the --with-shared-lib option.
The fix is to apply the following patch to $CCP4/configure:
diff -r1.165 configure
630,631c630,634
< SHARED_LIB_FLAGS=${SHARED_LIB_FLAGS-'-n32 -mips3'}
< SHARE_LIB=${SHARE_LIB-'ld -n32 -mips3 -shared -all libccp4.a -o libccp4.so'}
---
> # Shared libraries only if requested
> if test "$shared_lib" = yes; then
> SHARED_LIB_FLAGS=${SHARED_LIB_FLAGS-'-n32 -mips3'}
> SHARE_LIB=${SHARE_LIB-'ld -n32 -mips3 -shared -all libccp4.a -o libccp4.so'}
> fi
and then rebuild by running config.status, then make etc.
NB. you will need to delete the $CLIBS/libccp4.so and $CCP4_LIB/libccp4.so files before rebuilding.
We have had several reports of compilation failing on IRIX with Error Code 127 (bu21), or Error Code 2 (bu21). Sometimes this will be accompanied with TMPDIR undefined. The solution is to set TMPDIR during the configuration, eg
configure irix64 --with-x --tmpdir=/tmp/XXX(Thanks to Jonathon Dill)
Several FFLAGS and CFLAGS are missing from the x-windows/Mosflm/mosflm Makefile.in . This results in mosflm failing to compile for irix64 due to several of the produced objects being -n32. Replace x-windows/Mosflm/mosflm/Makefile.in with the version on the prerelease area.
If make has already been run; delete the *.o files in x-windows/Mosflm/mosflm, then execute config.status in x-windows/Mosflm, followed by make in x-windows/Mosflm/mosflm .
When the main CCP4 configure tries to set up to build the MapSlicer Tcl command library, it may fail to find an appropriate Tk library even though one exists on the system.
This is due to a bug in the $CCP4/configure, and can be addressed by applying the following patch:
diff -r1.171 configure 2215,2216c2215,2216 < # ... and possible versions for tk < for testvers in $tryvers --- > # ... and possible extensions for tk > for testext in $tryexts(Thanks to Michael Strickler)
There is a buffer size problem which manifests itself as a fatal error on attempting to write a scratch file, if there are 5 or more datasets. The fix is to $CPROG/scala_/dts_storage.fh:
diff ../scala_4.2/dts_storage.fh . 20c20 < common /dts_storage/ dts_data(maxsets*4+4), --- > common /dts_storage/ dts_data(maxsets*5+4),(Thanks to Phil Evans.)
Mosflm no longer handles R3 or R32 (hexagonal setting) properly because of the changes in the symop.lib file in CCP4 4.2. (H3 and H32 aren't handled in any release version of Mosflm.)
Anyone who has an MTZ file for R3 or R32 created more than a couple of weeks ago will find that the symmetry is broken, and they have to use MTZUTILS or CAD to change the space group as appropriate.
(Harry Powell)Although the documentation suggests that it possible to run AstexViewer as a standalone application, there are problems doing this - you get an error of the form
Exception in thread "main" java.lang.NoClassDefFoundError: MoleculeViewer
The problem is fixed by picking up a new version of the AstexViewer.jar file from the prerelease area. This replaces the version in $CPROG (and $CBIN, if you have already performed the installation).
(Thanks to Mike Hartshorn)anglen fails with a segv or gives incorrect results. This is corrected using the following patch which reinstates the presistance of data for calls of STORAT and STORPL.
diff -c -2 -r1.9 anglen.f
*** anglen.f 2002/02/25 10:04:32 1.9
--- anglen.f 2002/06/28 13:43:51
***************
*** 742,745 ****
--- 742,747 ----
SUBROUTINE STORAT(IRESID,RSNAME,INCHN,INSEQ,IBAT,COORDS,NEWCHN)
+ SAVE CHAIN, FILLED, XYZ, RNAME, SEQNO, SRESID
+
INCLUDE 'anglen.inc'
***************
*** 1076,1079 ****
--- 1078,1083 ----
SUBROUTINE STORPL(IRESID,RSNAME,ATNAM,INCHN,INSEQ,COORDS,NEWCHN)
+ SAVE CHAIN, NGOT, FILLED, XYZ, SEQNO, SNAME, SRESID, PLRES
+
INCLUDE 'anglen.inc'
Also for bplot.f, subroutine REAPLN
diff -c -2 -r1.5 -r1.6
*** bplot.f 2002/02/25 10:04:32 1.5
--- bplot.f 2002/07/02 08:42:43 1.6
**********************
*** 2638,2641 ****
--- 2641,2646 ----
SUBROUTINE REAPLN(LINE,RMS,LMAIN,LSTCHN,LSTRES,LSTSEQ,FOUND,
- COORD,ANAME,NATOM,ENDFIL)
+
+ SAVE IREC, IATOM
INCLUDE 'bplot.inc'
(Thanks to Fukami Takaaki)
There are a few cases of Refmac 5.1.19 (distributed with CCP4 4.2) terminating with the error:
Too many atoms in one brick. Maximum allowed in one brick = 60 Check symmetry, cell dimensions ===> Error: CREATE_NET failedI'm not sure what the cause of this problem is, but it appears to have been cured in Refmac 5.1.24 (distributed with CCP4 4.2.1). The advice is therefore to upgrade!
However be aware that the message may be genuine. One user reports inserting sequence with mut_ins in O, but not building them. Thus, there were greater than 60 atoms at 0, 0, 0 and refmac5 correctly complained!
(Martyn Winn)
The mapcutting task fails due to a broken .def file, apply the following patch to $CCP4I_TOP/tasks/mapcutting.def:
*** mapcutting.def 2002/03/07 10:03:12 1.1 --- mapcutting.def 2002/05/01 15:02:33 *************** *** 43,48 **** --- 43,49 ---- NCS_OP_XTRAN,0 _real "" NCS_OP_YTRAN,0 _real "" NCS_OP_ZTRAN,0 _real "" + NCS_OP_CONVENTION,1 _text POLAR NCS_OP_ALPHA,1 _real 0.0 NCS_OP_BETA,1 _real 0.0 NCS_OP_GAMMA,1 _real 0.0(Thanks to Kevin Cowtan)
There are two reported problems:
Error: unknown option "-dash"when stepping through bond types by clicking on the bond.
The task fails with ERROR no type for ANOMALOUS. This is corrected by appling the following patch to $CCP4I_TOP/tasks/import.def:
*** import.def 2002/01/30 15:17:41 1.5 --- import.def 2002/05/08 10:54:59 *************** *** 70,75 **** --- 70,76 ---- FALLOFF _logical 1 RUN_UNIQUE _logical 1 + ANOMALOUS _logical 1 UNIQUE_INPUT _MTZ_file "" UNIQUE_OUT _MTZ_file "" FREER_MODE _freer_mode 2
If the "Plot default Harker map sections with coordinates of peaks in map" option is chosen for orthorhombic spacegroups, then the task fails with an error message from MSYMLB3. This is basically because the Patterson spacegroup occupies more space in the CRYST1 line of the PDB file than is allowed for. There is a fix to the library file $CCP4/lib/src/rwbrook.f to deal with this case.
Download the fixed $CCP4/lib/src/rwbrook.f file from the prerelease area, and re-compile the CCP4 library. If you have compiled statically, then you will need to re-compile at least the peakmax executable against the fixed library.
When running lsqkab through ccp4i (superpose) the requested rms table will not display under loggraph. This is due to a double processing of the output table, which results in two headers. A quick fix is to remove the second header. A version of superpose.script which fixes this problem is on the prerelease area.
When running with NCS averaging and the option to Create file to display NCS operators in [...] format, the options are reversed i.e. requesting "XtalView" results in an "O" macro and vice versa.
The following patch applied to $CCP4/ccp4i/tasks/dm.tcl fixes the problem:
diff -r1.7 dm.tcl
39c39
< set typedef(_dm_vuout_format) { menu { O XtalView } { vu odat } }
---
> set typedef(_dm_vuout_format) { menu { O XtalView } { odat vu } }
(Thanks to Kevin Cowtan)
When running in "review" mode, the interface may complain that no harvest information is set even though there is no where in the window to enter it.
The following patch applied to $CCP4/ccp4i/tasks/refmac5.tcl will fix this behaviour:
diff -r1.20 -r1.21
65,66c65,66
< # Check harvesting if not running idealisation (and harvesting is selected)
< if { ![StringSame $refine_type IDEA] && \
---
> # Check harvesting if not running idealisation or review (and harvesting is selected)
> if { ![StringSame $refine_type IDEA REVIEW] && \
This affects all versions of CCP4i - in certain circumstances (e.g. trying to open a task window using the "rerun job" option), you get a Tcl error dialogue window with a message of the form:
can't set "array(REFINE_TYPE)": bad window path name ".w_dbtmp_2.main.canvas.contents.param_1"and the task window doesn't appear.
This seems to happens when the task window has previously been closed using the window manager control (e.g. the "X" close button on the top right of the window frame on Linux systems) rather than the "Close" button at the bottom of the window.
The errant behaviour is fixed by applying the following patch to $CCP4/ccp4i/src/taskwindow.tcl:
diff -r1.8 -r1.9 166d165 < 377a377,388 > # Bind destroy event to be consistent with the use of the > # Dismiss button above > # NB > # You can't just do "bind $w <Destroy> "..."" because all the > # children of $w also inherit the binding - so this construct > # (borrowed from Harrison and McLennan's "Effective Tcl/Tk > # Programming") defines a new bindtag and gives this to $w only > bind destroyTaskWindow <Destroy> \ > "UnsetArrayExtras $arrayname > DeleteTaskWindow $w $arrayname $taskname" > set tags [bindtags $w] > bindtags $w [linsert $tags 0 destroyTaskWindow](or by avoiding use of the window manager close control). (Thanks to Kevin Cowtan)
SCALA now uses information about datasets in the input MTZ file: if multiple datasets are present (from multiple runs of MOSFLM), the default is to scale tham together but split them into separate output files (this is intended for MAD datasets). In MOSFLM, if the user doesn't specify project and dataset names for runs, then the program gives each run a unique project and dataset name based on the date and time.
The problem occurs when these runs are sorted together and put into SCALA. By default SCALA treats each dataset independently, and outputs two MTZ files with names based on the dataset names from the input file: this is confusing if you wanted to just scale all the data together.
The ounce of prevention is to make sure that you explicitly set appropriate project and dataset names when running MOSFLM, using the PNAME & DNAME commands.
The pound of cure (workaround) using the existing SCALA gui is to define separate runs (the default "run 1 all" is OK for this purpose) and assign this run to a single output project/dataset (ie delete all but one dataset & give it a sensible name). Alternatively REBATCH can be used to reset the project and dataset names before running (but be aware that this options are not currently available through the REBATCH gui and must be scripted manually). If you are running SCALA with a script, add a command line
name project <project_name> dataset <dataset_name>to force all data into a single dataset.
An updated version of the SCALA interface has a new option to combine all input datasets into a single output dataset, which should solve the problem - this can be obtained from the prerelease area and requires the files $CCP4/ccp4i/tasks/scala.tcl, $CCP4/ccp4i/tasks/scala.def, $CCP4/ccp4i/templates/scala.com
(Thanks to Phil Evans)A second problem is that when using "Rerun" for these jobs, the details of output datasets (and of which run is assigned to which dataset) seem to be lost or overwritten.
Both these are fixed by an updated version of the SCALA interface file $CCP4/ccp4i/tasks/scala.tcl, which can be
Two problems:
Both problems can be fixed by replacing the existing $CCP4/ccp4i/src/local.tcl file with the version on the prerelease area.
The @filename instruction in the parser only uses the first 61 characters of the string. This means long filenames cannot be used. The solution is to increase the dimension of the CHARACTER array used to store the filename in parser.f.
169c169 < CHARACTER FLNAME*60,LINEX*1500,LINEK1,SLASH --- > CHARACTER FLNAME*256,LINEX*1500,LINEK1,SLASH
In ccp4ccif_setup_context a passed fortran character string is given a fixed length. This is bad coding practice. As additional memory may be grabbed for the string, leading to undefined behaviour.
This may have manifested itself in some cif2mtz problems, eg
>>>> CCIF signal CCIF_NOITEM (severity: Warning) <<<<
(Raised in ccif_setup_context)
Data name or category "REFLNFound column with label and type Column " not defined in dictionary
Apply the following correction to cciflib.f:
diff -r1.6 cciflib.f 3153,3154c3153,3154 < character*(cfllen) catnam_in,catnam_out < character*(*) disposition --- > character*(cfllen) catnam_out > character*(*) disposition, catnam_inUsers of the GNU compiler set will be required to make the following additional changes:
3166,3168c3166,3168 < write(ERRLINE,*) < + ' Warning in ccp4ccif_setup_context: Category '// < + catnam_in(1:lenstr(catnam_in))//' not present in data block.' --- > write(ERRLINE,"(A,A,A)") > + ' Warning in ccp4ccif_setup_context: Category ', > + catnam_in,' not present in data block.' 3175,3176c3175,3177 < write(ERRLINE,*) ' Error in ccp4ccif_setup_context: '// < + 'Tried to set up category '//catnam_in(1:lenstr(catnam_in))// --- > write(ERRLINE,"(A,A,A)") > + ' Error in ccp4ccif_setup_context: '// > + 'Tried to set up category ',catnam_in, 3184,3185c3185,3187 < write(ERRLINE,*) ' Error in ccp4ccif_setup_context: '// < + 'Tried to set up category '//catnam_in(1:lenstr(catnam_in))// --- > write(ERRLINE,"(A,A,A)") > + ' Error in ccp4ccif_setup_context: '// > + 'Tried to set up category ',catnam_in, 3191,3193c3193,3196 < write(ERRLINE,*) ' Error in ccp4ccif_setup_context: '// < + 'Unexpected context type for category '// < + catnam_in(1:lenstr(catnam_in)) --- > write(ERRLINE,"(A,A)") > + ' Error in ccp4ccif_setup_context: '// > + 'Unexpected context type for category ', > + catnam_in(Thanks to Avi Naim)
The documentation for "Implementing Data Harvesting" is missing from the programmers documentation for CCP4i.
The file $CCP4/ccp4i/help/programmers/harvest.html can be fetched from the prerelease area.
(Thanks to Maria Turkenburg)