next up previous contents
Next: Interacting with the program Up: A do-er's guide Previous: In more detail :   Contents

In more detail : Running the program

In the directory where the data.hkl and model.pdb (or model?.pdb) files are --and assuming that the symbol Qs has been defined (as discussed in the installation section), give : 11Qs -auto <xxx> where <xxx> is the total number of search models per asymmetric unit of the target structure, or Qs -reso <low> <high> -auto <xxx> if you want to explicitly define resolution limits (for example, Qs -auto 1 or Qs -reso 20 4 -auto 2, etc.).12The program will do its best to determine parameters suitable for the given problem, and if it succeeds, it will start the calculation. The name of the game in this procedure is memory : if the physical memory of your system is adequate for storing the whole molecular transform (or transforms in the multi-model case), then you can stop the program and submit it as a batch job. If, on the other hand, QS allocated 128 MBytes of memory when only 64 MBytes of it is physical memory, the program will never start : it will endlessly be reading and writing to the swap space on the disk with no chance of even finishing the FFT step. If that's the case, you will probably have to reduce the resolution of your data and try again. What I suggest you do, is to run the program interactively till it gets past the FFT stage. For medium-sized proteins and with adequate physical memory, this should only take a few minutes (or even seconds). If on the other hand the physical memory is not enough, the program will get stuck in the electron density calculation and FFT steps. What you should see (in the case of a multi-model problem) should look similar to this :

                                                                          
host# Qs -reso 50 4 -auto 3

            QQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQ          
            QQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQ          
            QQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQ          
             QQQQQQQQQQQQQQQQQQQQQQ QQQQQQ QQQQQQQQQQQQQQQQQQQQQQ           
             QQQQQQQQQQQQQQQQQQQQQ   QQQQ   QQQQQQQQQQQQQQQQQQQQQ           
              QQQQQQQQQQQQQQQQQQQ    QQQQ    QQQQQQQQQQQQQQQQQQQ            
               QQQQQQQQQQQQQQQQQ     QQQQ     QQQQQQQQQQQQQQQQQ             
                 QQQQQQQQQQQQQ      QQQQQQ      QQQQQQQQQQQQQ               
                    QQQQQQ          QQQQQQ          QQQQQQ                  
                                   QQQQQQQQ                                 
                                  QQQQQQQQQQ                                
                                 QQQQQQQQQQQQ                               
                                 QQQQQQQQQQQQ                               
                                  QQQQQQQQQQ                                
                                    QQQQQQ                                  
                                                                            

                                Queen of Spades                             
                                  Version 1.3                               
                                                                            

_______________________________________________________________________________

############################################################
##                                                        ##
##              Automatically created by Qs               ##
##                                                        ##
############################################################

############################################################
#
# Target function (can be R-FACTOR, CORR-1 or CORR-2) and
# number of minimisations and steps.
#
TARGET          CORR-1
CYCLES          5
STEPS           30000000

############################################################
#
# Annealing schedule & move size control.
# This is for a Boltzmann annealing schedule with a constant
# move-size and a fixed starting temperature.
#
BOLTZMANN
STARTING_TEMP   0.0700

############################################################
#
# Reflection selection.
#
KEEP            0.70
AMPLIT_CUTOFF   1.0
SIGMA_CUTOFF    2.0
RESOLUTION      50.00 4.00
RANDOM_SELECT   1.0

############################################################
#
# Fraction of reflections for free set.
#
FREE            0.10

############################################################
#
# Files to use for reading model(s) and data.
#
MOD1            model1.pdb
MOD2            model2.pdb
MOD3            model3.pdb
DATA            data.hkl

############################################################
#
# Scales, grid, random number generator seed, B-factors etc.
#
GRAUTO
SCMODE          wilson
INTERPOLATION   linear
SEED            47579
B_M1            20.00
B_M2            20.00
B_M3            20.00

############################################################
#
# Log file and postscript-related things.
#
INFO            1000
POSTSCRIPT      colour

############################################################
#
# Finally, cell, space group and molecules per a.u.
#
CELL             64.660  85.460  83.370 90.000 112.030 90.000
GROUP           4
MOLECULES       3
#
#
############################################################


_______________________________________________________________________________

Minimisation performed against the 1-Corr(Fo,Fc) target.
Will perform 5 independent minimisations.
Number of steps for minimisation : 30000000
Fraction of strongest reflections to keep : 0.700000
Amplitude cutoff set to    1.00
Reflections with F/sigma(F) less than  2.00, will be rejected.
Resolution limits set to   50.00 -  4.00 Angstrom.
Only a fraction  1.000 of the available reflections will be used.
Free value will be calculated over a fraction 0.100 of the data set.
Model number 1 : PDB file name set to model1.pdb
Model number 2 : PDB file name set to model2.pdb
Model number 3 : PDB file name set to model3.pdb
Reflection file name set to data.hkl
Will try to determine SCALE and MAXGRIDSPACING automatically.
Wilson-like scaling will be used.
Will be using linear interpolation.
Random number generator reset to 47579
Model number 1 : overall B-factor set to 20.000000
Model number 2 : overall B-factor set to 20.000000
Model number 3 : overall B-factor set to 20.000000
Information about the minimisation will be printed every 1000 moves.
Colour postscript output requested.
Cell parameters  64.66  85.46  83.37  90.00  112.03  90.00
Space group    4 with   2 symmetry operators
Symmetry operator :
                 -1 +0 +0
                 +0 +1 +0
                 +0 +0 -1
with translation vector : +0.0000000 +0.5000000 +0.0000000
Number of molecules in asymmetric unit : 3
2 symmetry operators read in.
Space group given. Lattice type is P (only used for the packing diagrams).

Target function for minimisation : 1.0-Corr(Fo,Fc)
No bulk-solvent correction requested.

Temperature control :
         Boltzmann scaling selected : the temperature at each step (k)
         will be given by T=T0/ln(k). T0 is set to the value defined by
         the keyword STARTing_temp.
Move size control :
         The modulus of the moves attempted is constant and independent of
         either time or R-factor (or correlation). Its maximum value only 
         depends on the resolution of the input data.

_______________________________________________________________________________


Reading atoms (Model 1)                 2240 atoms read (0 unknown)
Centre of mass at                         19.749   23.683   64.488
Box dimensions (A) :                      57.433   58.288   56.652 along x,y,z
Translating/rotating ...                done.
Centre of mass at                          0.000    0.000   -0.000
Box dimensions (A) :                      65.001   51.063   50.244 along x,y,z

Reading atoms (Model 2)                 869 atoms read (0 unknown)
Centre of mass at                         39.417   33.807   65.919
Box dimensions (A) :                      49.416   32.517   81.659 along x,y,z
Translating/rotating ...                done.
Centre of mass at                          0.000    0.000   -0.000
Box dimensions (A) :                      85.850   41.229   27.084 along x,y,z

Reading atoms (Model 3)                 814 atoms read(0 unknown,38 P->S)
Few or no CA atoms. Packing diagrams will include all atoms (keyword PACKall).
Centre of mass at                         37.756   36.302   41.899
Box dimensions (A) :                      75.250   22.857   24.875 along x,y,z
Translating/rotating ...                done.
Centre of mass at                         -0.000   -0.000    0.000
Box dimensions (A) :                      75.564   22.832   25.470 along x,y,z

Calculating |F| cutoff ...                112.70
Reading reflections ...                 4943 read.
Reflections for free value              525
Excluded reflections                    17136
Lowest resolution reflection :          19.5 Angstrom
Highest resolution reflection :         4.0 Angstrom
Scale set to                             4.000
Big cell :                               343.400  204.253  200.977
Grid                                     384  210  210
 with spacing                              0.894    0.973    0.957
Physical memory required                193 MBytes
Allocate memory ...                     done.
FFTW is learning how to do FFTs ...     done.
Saving FFTW's wisdom file ...           done.
Atomic density profiles (B=20.0) ...    done.
Make electron density map (Model 1) ... done.
Atomic density profiles (B=20.0) ...    done.
Make electron density map (Model 2) ... done.
Atomic density profiles (B=20.0) ...    done.
Make electron density map (Model 3) ... done.
Write out projections ...               done.
Calculate molecular transform (1) ...   done in 5 seconds.
Calculate molecular transform (2) ...   done in 5 seconds.
Calculate molecular transform (3) ...   done in 5 seconds.
Rearranging transforms  ...             done.
Write out central sections ...          done.
Initialisations ...                     done.
Ready to roll after ...                 72 seconds.

_______________________________________________________________________________

Starting minimisation   1.
Initial 1-Corr(Fo,Fc)  0.94847
Starting free value   0.88592
$TABLE: Qs simulation 1:
$GRAPHS
:1-C(Fo,Fc)        vs time:A:1,2:
:Free 1-C(Fo,Fc)   vs time:A:1,3:
:Temp              vs time:A:1,4:
:1-C(Fo,Fc) & Free vs time:A:1,2,3:
$$


           TIME      1-C(Fo,Fc)     FREE        TEMP
$$
$$
            1000      0.931011    0.921275    0.010132
            2000      0.921983    0.927676    0.009209
            3000      0.924896    0.916017    0.008743
            4000      0.922867    0.919347    0.008440
	    ..........................................

You can now stop the program, and submit a proper batch job : for unix something like

batch
/usr/local/bin/Qs -reso 50 4 -auto 3 > LOG
<CTRL-D>
and for VMS :
submit/noprint/notify/queue=for_ever/log=dsk$23:[my.dir]QS.LOG Qs.com
where the Qs.com looks like
$ set def dsk$21:[my.directory]
$
$ Qs :== $dsk$56:[Qs.directory]Qs-dec4000-vms6.exe
$ Qs -auto 1
$
$ exit



Footnotes

... give :11
Plenty of memory ?
If you have loads of physical memory, you can make use of this extra memory to speed-up the program by a factor of approximately 25%. This you do as follows : start the program as discussed in the main text, stop it with CTRL-C, edit the file Qs_auto.in, remove the line that says GRAUTO, add a line SCALE 8.0, add a line MAXGRID 1.0, and finally, change the line INTERPOLATION linear to INTERPOLATION none. Save the modified Qs_auto.in file and give Qsi Qs_auto.in. The program will probably stop with an error message about how much memory is required with this set of parameters. If you have enough memory, use Qsi -force Qs_auto.in to run the program (the -force flag is needed both for interactive and batch jobs).
... etc.).12
If you have defined a symbol Qsi pointing to the QS executable (as discussed in the installation section), try running the program interactively using the Qsi instead of the Qs symbol. The reason is that possible error messages will be shown in bold, and are thus significantly easier to spot.

next up previous contents
Next: Interacting with the program Up: A do-er's guide Previous: In more detail :   Contents
NMG, January 2005