Q. I used the online form a few weeks ago to obtain ftp
instructions and a password. I want to download everything again but
the password doesn't work. Do I have to ask you for the new password?
A. No. You should fill out the online form again. This form can
be filled out at any time to obtain the current ftp password (it
changes automatically every week).
Q. I have downloaded the CNS distribution but when I try to
extract it I have problems.
A. Use the following to unzip and untar the distribution:
gunzip -c file-name | tar xvf -Check that you have enough disk space to accommodate the unpacked files (you will need at least 150 MBytes).
Q. I successfully unpacked the CNS distribution but when I try to compile I get an error:
testing Fortran and C compilers compiling: cc -mips4 -n32 -DCNS_ARCH_TYPE_SGI C compiler passes test compiling: f77 -DCNS_SOLVE_COMPILE -woff all -O2 -WK,-o=1,-so=0 -OPT:roundoff=2: IEEE_arithmetic=3:fold_arith_limit=2000 -mips4 -n32 f77 ERROR: cannot exec /usr/lib32/cmplrs/mfef77 *** Error code 2 (bu21) *** Error code 1 /usr/sbin/smake: Error: 1 error *** Error code 2 (bu21) compiler problems - stopping installation please check compilers before retrying installation *** Error code 1 (bu21)A. This error message is produced by the compiler tests. These tests are very simple. If something goes wrong, it is almost certain that your compiler installation is broken. For some reason this happens very often on Silcon Graphics Systems. The only advise we can give is that you re-install the compilers.
Q. I successfully compiled CNS on an SGI, but when I try to run it on a different SGI the program doesn't even start and I get an error:
rld: Fatal Error: cannot successfully map soname ....A. Make sure that the same libraries are installed on both machines. In particular you should look out for the libfastm, libcomplib and libblas libraries. In many cases different libraries are installed by default depending on the version of Irix and the hardware type.
Q. I successfully compiled CNS on an SGI O2 running Irix 6.3. However,
when I try to run the program it stops with a message about "Test of
FFT3C failed".
A. This is due to a bug in the SGI complib libraries which can
be observed under Irix 6.2, 6.3 and 6.4 (depending on the compiler
version). To recompile CNS without the complib libraries:
source cns_solve_env make fftfile-generic setenv CNS_FFTLIB "" make cns_solveIf the FFT test fails under both the SGI complib libraries and the generic FFTPACK code then this probably indicates a compiler problem. We suggest upgrading to the latest version of the Fortran compiler and making sure that the appropriate SGI patches are installed.
Q. I am trying to run CNS on SGIs with Irix 6.2, 6.3 and 6.4 with
a variety of different patches applied to all the machines. I am
having problems either compiling or running the program.
A. Upgrade to the most recent version of Irix 6.5 and install
and SGI recommended patches.
Q. How do I compile CNS with 64-bit addressing on an SGI (the
-64 compiler option)?
A. In general this is not a useful thing to do as there is no
gain in performance. However, if a structure is so large that an
address space larger than 2GB is required then this compilation may be
needed. To recompile the program:
source cns_solve_env make clean setenv CNS_ABI '-64' make cns_solve
Q. I am trying to compile on a Linux machine running RedHat version 6.0. I installed the latest f2c RPM so I could use fort77 for the compilation. However, when the CNS installation stops:
testing Fortran and C compilers compiling: cc -O -ffast-math -DCNS_ARCH_TYPE_LINUX C compiler passes test compiling: fort77 -w -Nn2000 -O3 -malign-double -funroll-loops -ffast-math MAIN hello: linking: fort77 ***** ERROR: problem with Fortran compiler ***** make[3]: *** [fortran-test] Error 2 make[2]: *** [compiler-test] Error 2 make[1]: *** [compiler-test] Error 2 compiler problems - stopping installation please check compilers before retrying installation make: *** [install] Error 1A. The f2c libraries need to be recompiled (otherwise any code linked with them will produce a segmentation fault when run). We suggest obtaining the latest version of f2c from NETLIB and compiling the libraries from scratch.
Q. I've heard that there are optimized FFT libraries (DXML)
available for DEC Unix. How do I compile CNS with these DXML libraries?
A. In our experience the DXML libraries result in slower FFT
code (the FFTPACK code distributed with CNS is faster for the majority
of cases and is therefore used by default). However, to compile with
the DXML libraries:
source cns_solve_env setenv CNS_DEC_DXML "" make fftfile make cns_solve
Q. I tried to compile the utility programs by hand but they
didn't compile.
A. The utility programs are compiled automatically at install
time. They should be in your path (ie. you can just type the name of
the program at the Unix command line). If you want to compile them
again type: make utils
Q. When I start up CNS on my SGI it stops with the following message:
Fatal Error: test of numerical function failedWhat does this mean and how can I fix it?
make static-fastmIf this still produces the fatal error message then you should compile without the fast math library.
make no-fastmThis problem is present in both Irix 6.5.9 and 6.5.10.
Q. I am running on an SGI and I get results which are very
different from any of the other types of machines I use (even when I
run exactly the same job). Is there a problem? If so what should I do?
A. First keep in mind that because of numerical differences
between computing platforms there will sometimes be small differences
in output. However, if the differences are large it is possible that
the cause is a bug in the fast math library on Irix6.5. This problem
has been seen on R5000 machines with the 7.2 and 7.3 versions of the
compilers. When the dynamically linked fast math library is included
in compilation the program produces incorrect results for some
operations. To overcome this problem you should first try recompiling
with the static version of the library:
make static-fastmIf this still produces the same incorrect results then you should compile without the fast math library.
make no-fastmThis problem is present in both Irix 6.5.9 and 6.5.10.
Q. I am running on a Linux machine with RedHat 7 and I get
results which are very different from any of the other types of
machines I use (even when I run exactly the same job). Sometimes jobs
just crashes with a strange error message about coding errors. Is
there a problem? If so what should I do?
A. It appears that there may be problems with the Fortran/C
compilers distributed with RedHat 7. The RedHat developers choose to
distribute the development version of GCC (2.96), which is not
supported by the GCC developers. Problems have been reported when
using the CCP4 program REFMAC under RedHat 7. These problems have been
traced to the compiler version. See Kevin Cowtan's page for
further information. We recommend:
Q. The CNS HTML interface seems really slow, I am located in Europe
and have a slow network connection.
A. By default the interface will access the web server at Yale. If you
run your own local web server then you can install a local version of
the CNS HTML interface. Please check the CNS installation notes for
instructions.
Q. I installed the CNS HTML interface locally on our web server
but it doesn't run. I get an error about the PERL version when I try
to use the interface.
A. Make sure that you are using PERL version 5. The CGI scripts
will not work with PERL version 4. The latest version of PERL can be
obtained from the PERL
website.
Q. Where is the CNS manual?
A. The documentation is all HTML based. There are several
different sources of information:
Q. I ran a CNS task file but I think it may have failed, how
can I check this?
A. The first thing to check is that the task file terminated
without an error. On Unix systems this can be done by tailing
the last 30 lines of the output file (for example, tail -30
job.out). If there is no obvious error then on Unix systems the
output file can be searched for errors using the grep command
(for example, grep ERR job.out). If an error is found the
context may not be obvious. The job should then be rerun after setting
the log_level variable (located in the body of most task file) to
verbose:
evaluate ($log_level=verbose)
Q. I am trying to run torsion angle dynamics and I get an error
when it tries to generate the torsion angle topology. It suggests that
I need to change some parameters, what do these parameters mean?
A. The covalent topology of the system must be converted into
something suitable for performing torsion angle dynamics
calculations. It is at this stage that problems may occur. The torsion
angle topology has the following hierarchy (from largest to
smallest):
Tree (a connected set of chains) |
Chain (an unbranched connected set of groups) |
Group (set of connected atoms with no rotating internal torsions) |
{* maximum number of distinct bodies *} {===>} torsion_maxtree=5;Alternatively the parameter MAXTREE can be increased in the dynamics torsion topology statement to overcome this problem.
{* maximum number of chains (increase for large molecules) *} {===>} torsion_maxchain=1000;Alternatively the parameter MAXCHN can be increased in the dynamics torsion topology statement to overcome this problem.
{* increase for long stretches of polyalanine *} {===>} torsion_maxlength=50;Alternatively the parameter MAXLEN can be increased in the dynamics torsion topology statement to overcome this problem.
Q. I have a ligand molecule in my structure that is a long unbranched chain (HOCH2CH2CH2CH2OH) without any hydrogen atoms included. When I try to run torsion angle dynamics the program stops with an error:
ERROR: There are no suitable base groups. This problem can be caused by isolated bonding networks with undefined or weak dihedral force constants.A. The problem is that the torsion angle algorithm needs to have a rigid group of bonded atoms which contains at least 3 atoms. In the case of the long linear molecule described above there is no such group. The following solutions are possible:
Q. I am trying to select atoms for NCS restraints but the program
seems to be ignoring the selection.
A. Do the segid names have lower case letters? It is important to tell
CNS to maintain the case by the use of quotes. If the segids are Mol1
and Mol2 then the selection would be: (segid "Mol1") and (segid
"Mol2"). In general we recommend that uppercase be used for all names
to avoid any case problems.
Q. I have a model with protein and water molecules and a couple
of ions. When I run the generate_easy script the output has warnings
about no matching FIRST and LAST patch and no matching LINK. Did I do
something wrong?
A. No. The generate_easy script attempts to interpret the model
from a single PDB file. The program tries to find out if there are any
patches that need to be applied between each residue sequential
residue, and at the beginning/end of each chain. If there are no patch
found (ie there is no patch to link water molecules together) it
reports this. This is information rather than an error. If you use the
generate.inp task file, where the input coordinates are divided into
separate PDB files, these warning messages are not produced.
Q. I have a large structure, when I run try to generate the
molecular topology I get mesages about some parameters being exceded
and that I need to recompile the program. This doesn't seem right
because I compiled CNS for more atoms than I have in my structure.
A. This is probably happening because the number of atoms in
your structure plus the number of hydrogens, which are added initially
during the moleculer topology generation, is greater than the limit
CNS has been compiled for. You can either recompile CNS for even more
atoms or you can use the protein_nohydrogen.top file for the protein
topology. As the name implies, this has no hydrogens.
Q. I have a molecule with lots of different substrates, and I
don't have any free slot left to enter all of them in the generate.inp
task file. How should I change the generate.inp in order to put all of
the substrates?
A. you have to options depending on your circumstances. If you
can keep all of your coordinates in one PDB file then you should be
able to use the generate_easy.inp task file. Alternatively if you have
to use the generate.inp task file you can add more entries for ligands
(or any other type) by duplicating the input fields and changing the
numerical extension (e.g. below another ligand input section has been
added with 2 replacing 1:
{* ligand coordinate file *} {===>} lig_coordinate_infile_2=""; {* rename segid *} {+ choice: true false +} {===>} lig_rename_2=false; {* new segid *} {===>} lig_segid_2=""; {* convert chainid to segid if chainid is non-blank *} {+ choice: true false +} {===>} lig_convert_2=false;
Q. I located heavy atom sites automatically from my MAD data
and I am sure they are correct. However, the maps from the MAD phasing
look really bad, with no obvious solvent boundary.
A. The position of the sites is ambiguous. You will have to
calculate phases using the original site positions, and then repeat
this using the inverted coordinates (there is a CNS task file called
flip_sites.inp which can be used to "flip" (invert) the coordinates
for all the sites). Compare the 2 electron density maps to determine
which is correct, there should be a clear solvent boundary and
macromolecular features for the correct solution.
Q. I have data in an enantiomorphic space group: P6(2) or
P6(4). I know I have to do the phasing with the sites as found and
also with the sites inverted (flipped). Does this mean I have to run 4
phasing jobs (flipped and unflipped for both spacegroup choices)?
A. No, you only have to run 2 jobs if you have more than one
site. When inverting the sites you should also change the spacegroup
to the other enantiomer, ie. run one phasing job with unflipped sites
and spacegroup P6(2), and the other phasing job with flipped sites and
spacegroup P6(4). If you only one site you should probably run all 4
jobs. The flip_sites.inp task file will automatically change the
spacegroup in the output site database file if you have an
enantiomorphic space group, you must make sure the spacegroup in the
heavy atom phasing input file matches this.
Q. I think that one of the parameters in DENZO was set
incorrectly such that the Friedel mates (F+ and F-) in my MAD
experiment are switched. Do I need to go back to DENZO to change this?
A. No. You can just flip the Friedel mates in CNS. This can be
done with the flip_friedels.inp task file which will flip all
anomalous information in a reflection file (including phases).
Q. I have a partial molecular replacement solution for my data
and also a single derivative. Should I bother trying to calculate
phases from the derivative?
A. Yes. Any experimental phase information you can obtain will
make model building (using phase combined maps) and refinement (using
the MLHL target) easier. Remember that the experimental phases will be
as unbiased as you can hope for (the molecular replacement phases will
not be). You could use the phases from the partial molecular
replacement model as prior phase information in your SIR phasing.
Q. I have carried out MAD phasing in CNS but I want to
calculate the experimental electron density map with a different
program. The other program will not accept anomalous data how can I
convert the CNS MAD phasing results to non-anomalous?
A. The CNS task file average_friedels.inp can be used for this
purpose. It will convert the amplitude and phase information such that
the anomalous signal is averaged. This information is written out in a
new CNS reflection file. This is not really appropriate for cases
where there is a very large anomalous signal.
Q. I want to use my density modified phases in the refinement
of my structure. I am worried that the significance of the phase
probabilty distributions might be overestimated can I do anything
about this?
A. Yes. You can "blur" the phase probabilty distributions by
application of a scale factor and B-factor. The CNS task file
hlcoeff_blur.inp should be used for this. However, in general it is
always best to use the raw experimental phases in refinement (tests
indicate that this will provide the best model). Inflated FOMs are
a particular problem for density modified SAD phases, in this case the
raw experimental phases should always be used.
Q. I have carried out MAD phasing using some borderline
data. The maps don't look to good and the refined f' and f'' values
(in the .fp file) are not what I expected. In particular the f' values
are positive for one set of LOC expressions but negative for
another. What should I do?
A. Although it is difficult to give a general answer to the
cause of the problem, most likely your dispersive differences have
large errors. This can result from problems in data collection or more
often from crystal decay (which is still a problem in many cases even
with cryo-cooling). If you are confident that your anomalous
differences look reasonable it is sensible to treat this as a SAD
experiment, using the wavelength with the largest anomalous signal. A
solvent content of 50% or higher will increase your chances of this
working. Please see the tutorial section for details of how to perform
SAD phasing. If you have multiple wavelengths with an anomalous signal
you can treat each as a separate SAD experiment then use phase
combination to generate an optimal set of phases for density
modification (the task file xtal_util/combine.inp should be used for
the phase combination).
Q. Is there a locked rotation or translation search available
in CNS?
A. No.
Q. I'd like to use an EM mask for molecular replacement to get
low resolution phases. How should I go about doing this?
A. You can use Gerard Kleywegt's Moleman2 program to fill a
mask with atoms. You can then use the single_atoms.top and
single_atoms.par files for generating the molecular topology
information. Note that the residue name and atom names in the PDB file
must match those in the topology and parameter files.
Q. What should I use for refinement, simulated annealing or energy
minimization?
A. Simulated annealing is most appropriate at the start of refinement
when the model has relatively large errors. Towards the end of
refinement when the model is essentially correct, or you are just
placing water molecules then energy minimization is more appropriate.
Q. What crystallographic target should I use for refinement?
A. Results indicate that maximum likelihood targets give better
results than traditional least-squares targets. In CNS there are 3
maximum likelihood targets: MLF (amplitudes), MLI (intensities) and
MLHL (amplitude and Hendrickson-Lattman coefficients). If experimental
phase information is available (even SIR/SAD phases) the MLHL target
should be used. If no phase information is available (a molecular
replacement solution) then either the MLF or MLI target should be
used. The MLI target can used experimental intensities (which are
written out by to_cns), if these are not present then the amplitudes
will be converted to intensities internally.
Q. How do I obtain a value for the weight between the
crystallographic and geometric terms (Wa)? I need to have this before
I start refinement, right?
A. The value of Wa is needed for refinement, but all CNS
refinement task files can calculate this weight internally - the user
does not have to specify the value (the value of Wa is entered as
-1). This is done because the value of Wa depends on the
crystallographic target and this in turn depends on the current
model. As the model improves, the weight changes. This is more
pronounced for the maximum likelihood targets which incorporate more
information about the current errors in the model into the target
function. If you have reason to think the automatically calculated
value for Wa is incorrect (this is not likely to be the case) a task
file is provided (xtal_refine/optimize_wa.inp) for carrying out a grid
search for the best value of Wa. Be aware that this takes some time to
run so should only be used if necessary. The optimum value of Wa can
then be used in the next few refinement steps, but should be updated
if the model changes significantly.
Q. Should I build hydrogens for refinement?
A. For most refinements at sub-atomic resolution (3.5 to 1.2
Angstroms) we feel that hydrogens are not necessary. Infact, it is not
possible to add all hydrogens with the default CNS crystallographic
topology/parameter files. These files only describe the polar
hydrogens not all hydrogen atoms. Electrostatic terms are not included
in crystallographic refinement, rendering these hydrogens unnecessary.
Q. My atomic B-factors are too high, why won't CNS make them lower?
A. The B-values are, in all likelihood, ok (what evidence do
you have that they are too high?). Older refinement methodologies may
have underestimated B-values by the use of restricted resolution
ranges and sigma cutoffs. By refining against all the data the
B-values are more realistic. Also, it has been suggested by some
researchers that the freezing of crystals may increase B-values.
Q. After individual atomic B-factor refinement the B-factors
are over- or under- restrained, is there anything I can do to change
this?
A. You should run the optimize_rweight.inp task file to obtain
a good estimate for the weight on the restraint term. This will
perform several B-factor refinement with different restraint weights.
This should find the optimal value of the restraint weight for
individual, restrained B-factor refinement.
Q. My data only extends to 3 Angstroms resolution, is it ok to
use the overall anisotropic B-factor correction in refinement?
A. Yes, this is an overall correction and at most uses 6
parameters. This is not the same as refinement of individual
anisotropic B-factors for each atom. The overall anisotropic
correction can be applied even when the upper resolution limit is 5
Angstroms or worse.
Q. Can I do individual anisotropic B-factor refinement in CNS?
A. No, this is not yet implemented.
Q. I picked water molecules then went back to refinement, but when I try to do some refinement the job crashes with:
%NBUPDA-ERR: missing nonbonded Lennard-Jones parameters %%%%%%% ATOM: SEGId="SOLV", RESId="1 ", NAME="OH2 ", CHEMical="OT "A. You forgot to include the parameter file for water molecules:
Q. How do I deal with alternate conformations in my refinement?
A. Unfortunately this is not very elegant at present:
Q. I have data which shows pseudo-merohedral twinning. Can I
refine my structure with CNS?
A. No. The twinned refinement in CNS is designed for use with
true hemihedral twinning and will not work correctly with data that
shows pseudo-twinning. You should use the program SHELXL instead. True
hemihedral twinning is possible in the following cases:
True point group | Twin operation | hkl related to |
3 | 2 along a,b | h,-h-k,-l |
2 along a*,b* | h+k,-k ,-l | |
2 along c | -h,-k , l | |
4 | 2 along a,b,a*,b* | h,-k ,-l |
6 | 2 along a,b,a*,b* | h,-h-k,-l |
321 | 2 along a*,b*,c | -h,-k , l |
312 | 2 along a,b,c | -h,-k , l |
23 | 4 along a,b,c | k,-h , l |
Q. How do I look at a CNS format electron density map in O?
A. Use MAPMAN (http://alpha2.bmc.uu.se/~gerard/manuals) to
convert the map file to O (DSN6) format. The CNS map file can be read
as X-PLOR format into MAPMAN.
Q. How do I look at an EZD format electron density map, written
by CNS, in O?
A. Use MAPMAN (http://alpha2.bmc.uu.se/~gerard/manuals) to
convert the map file to O (DSN6) format. The CNS EZD map file must be
read as NEWEZD format into MAPMAN.
Q. How do I convert a scalepack file directly to a CNS reflection file?
A. Use the to_cns tool provided with CNS:
to_cns file.sca file.hklThe output CNS reflection file will contain the original intensity information and values for the amplitudes and sigmas. Note that negative intensities will produce zero amplitudes. An alternative approach is to use TRUNCATE in CCP4 to generate amplitudes and then use mtz2various to generate the CNS reflection file.
Q. How do I convert a d*trek file directly to a CNS reflection file?
A. Use the to_cns tool provided with CNS:
to_cns -f dtrek file.ref file.hklNote that for anomalous data the d*trek reflection file should have the following format (which is the default in d*trek):
3 6 0 nH nK nL fIntensity fSigmaI fIntensity+ fSigmaI+ fIntensity- fSigmaI-The output CNS reflection file will contain the original intensity information and values for the amplitudes and sigmas. Note that negative intensities will produce zero amplitudes. An alternative approach is to use TRUNCATE in CCP4 to generate amplitudes and then use mtz2various to generate the CNS reflection file.
Q. How do I convert a CCP4 MTZ reflection file to CNS format?
In particular I want to get Hendrickson-Lattman coefficients
converted.
A. Bart Hazes program SFTOOLS (part of the CCP4 distribution)
can be used to read an MTZ file and write out a CNS reflection file
minus the header information (this can then be added manually). The
program mtz2various, in CCP4 v3.5, is able to perform a complete
conversion to CNS format. Example:
mtz2various HKLIN infile HKLOUT outfile << EOF RESOLUTION 10000 1 OUTPUT CNS EXCLUDE SIGP 0.001 # to exclude unmeasured refl. LABIN FP=Fnat SIGFP=SIGnat PHIB=PHI FOM=FOMM HLA=HLA HLB=HLB HLC=HLC HLD=HLD END EOF
Q. How can I convert a CCP4 MTZ file with anomalous data to a
CNS reflection file?
A. The program mtz2various, in CCP4 v3.5, is able to perform a
complete conversion to CNS format. If you have anomalous data as F and
deltaF:
# Convert MTZ file with anomalous data as F and deltaF to CNS format. # F(+h +k +l) and F(-h -k -l) are reconstructed from FP and DP. mtz2various hklin file.mtz HKLOUT file.hkl <<+ LABIN FP=F SIGFP=SIGF DP=DANO SIGDP=SIGDANO OUTPUT CNS END +If you have anomalous data as F(+) and F(-):
# Convert MTZ file with anomalous data as F(+) and F(-) to CNS format. mtz2various hklin file.mtz HKLOUT file.hkl <<+ LABIN F(+)=F(+) SIGF(+)=SIGF(+) F(-)=F(-) SIGF(-)=SIGF(-) OUTPUT CNS END +
Q. How does the CNS program treat atoms on special positions?
A. CNS will recognize atoms on special positions if they are
within 0.1 Angstroms of the position. This default can be changed:
parameter nbonds special_position=0.2 end endAtoms should be placed exactly on the special position if possible - if too far away they will be pushed off the position. Once recognized by CNS the atom will not interact with itself by crystallographic symmetry and its occupancy will be appropriately scaled by the multiplicity of the special position.
Q. How do the CNS refinement task files treat atoms on special
positions?
A. Atoms close enough to a special position will be recognized
as such and have their occupancy scaled and symmetry self-interactions
removed. In addition, the task files will fix these atoms so they do
not move in any direction during refinement (the position after
refinement will be the same as the starting position).
Q. Can I have an atom on a special position with other atoms
bonded to it (eg. a sulphate ion sitting on a 2-fold axis)?
A. No. CNS does not currently permit bonds to atoms on special
positions. There are some solutions to this:
Q. I have Zinc and Magnesium ions in my structure. The
refinement is going well, but when I look at a difference map (Fo-Fc)
I see positive density at the metal positions. Do I have a problem
with my structure?
A. The scattering factor library in CNS has different values
for the neutral (metallic) form of the element and the ionic (charged)
form. It appears that in some cases the electronic state of the
element may be closer to one or the other (probably depending on the
environment of the atom). It may be necessary to try using the
alternate atomic description and look at the electron density. In
order to do this the ion.top and ion.param files supplied with CNS
have residue definitions for both neutral and charged species (in some
cases more than one charged species is allowed). For example there are
2 residue definitions for Zinc:
ZN (with atom ZN) - no charge, will use the ZN scattering factor ZN2 (with atom ZN+2) - +2 charge, will use the ZN+2 scattering factorIn order to switch between the two, the PDB file needs to be edited to change the residue name and atom name for the Zinc atom prior to running the generate.inp task file.
Q. My structure contains a molecule which lies across a
symmetry operation. This means that some parts of the molecule are
mapped onto each other by symmetry. It is not a special position case,
because no one atom lies on the symmetry operation exactly. How can I
tell CNS to refine the molecule while allowing the overlap to
occur?
A. You can use the igroup statement to turn the
pvdw (Packing-Van-der-Waals) interactions off by setting the
weight to zero:
igroup interaction (molecule) (molecule) weights * 1.0 pvdw 0.0 end end
If you have other molecules in the asymmetric unit, you have to define the interactions explicitly. For example.
igroup interaction (rest) (rest) interaction (rest) (molecule) interaction (molecule) (molecule) weights * 1.0 pvdw 0.0 end end
Replace (rest) and (molecule) by the appropriate atom selections.
The next question is where to put the igroup statement. For example,
in minimize.inp or anneal.inp, search for
igroup. The statement you will find is for alternate
conformations. Right after the end statement of that igroup
block, insert your own igroup statements as outlined above.
This will only work if there are no alternate conformations. Otherwise
a more complicated sequence of igroup interactions must be
given, which simultaneously take both the alternate conformations and
the molecule across the symmetry operation into account.
Q. I am having difficulties using the input files
"generate.inp" and "generate_easy.inp" in version 0.5 of CNS. If the
input file is a PDB file of a protein with hydrogens, CNS creates mtf
and PDB files without protons even when the flag "hydrogen" is set to
true.
A. You also need to change the topology and parameter files to
those with all hydrogens:
{* protein topology file *} {===>} prot_topology_infile="CNS_TOPPAR:protein-allhdg.top"; {* protein parameter file *} {===>} prot_parameter_infile="CNS_TOPPAR:protein-allhdg.param";