RSEARCH (CCP4: Deprecated Program)

NAME

rsearch - R-factor and correlation coefficient between Fcalc and Fobs for positions of molecule within a specified grid.

SYNOPSIS

rsearch HKLIN foo_in.mtz [ SEARCHSAVE foo_search.tmp ] [ MAPOUT foo_rsearch.map ]
[Keyworded input]

DESCRIPTION

When a molecule is correctly orientated in a unit cell, but the translation parameters are not known, RSEARCH calculates the R-factor and correlation coefficient between calculated and observed structure factors for each position of the molecule within a specified grid.

You may prefer to calculate a translation function with TFFC instead of RSEARCH, since this is much faster.

There is an option to restart RSEARCH when it runs out of time : a file assigned to SEARCHSAVE will record the intermediate sums for the R factors. These are updated after every 100th reflection and a note is written to the log file. There are also options (keywords CENT and RZONE) to limit the reflections used to centrics or a specified zone.

KEYWORDED INPUT

The various data control lines are identified by keywords. Only the first 4 characters need be given. Those available are :

AXIS, BINMAP, BGRV, CENT, CORRELATION, END, FCLIMIT, FOLIMITS, LABIN, LPMAP, NRFACTOR, RESOLUTION, RESTART, RZONE, SCALE, SYMMETRY, TITLE, XGRID, YGRID, ZGRID

AXIS

Order of grid axes : fast, medium, slow.
At present, the program only looks at the slow (sectioning) axis. Possible inputs are therefore

                               - X Y Z
                               - Z X Y
                               - Y Z X

BINMAP

Output a map of R factors or correlation coefficients over the specified grid

BGRV <brgval>

Background R value for maps.
Default: the average R value.
The terms calculated are (background R value - grid R value).

CENT

By default, all reflections are included in the calculation. If this is set, ONLY centric reflections are selected. Centric zones are identified using the spacegroup operators.

CORRELATION

Turns ON the plotting of correlation maps

FCLIMIT <fmaxc>

Set lower limit on FC for inclusion in R factor calculation.

In low symmetry space groups such as P21, you should exclude all terms where one calculated (partial) structure factor FC is very small, since then the R factor will be independent of translation.

FOLIMITS <fmin> <fmax>

Set upper and lower limits on observed structure factors for inclusion in R factor calculation (given in either order).

LABIN <program label>=<file label>...

Input column assignments. The program labels are:

        H  K  L  FP  SIGFP  FC1  PC1 ... FC12  PC12  FC  PHIC

FP is the observed structure factor and SIGFP its standard error. FCn and PCn are the partial calculated structure factors and phases for symmetry operation n. Known structure may be added in by defining FC and PHIC.

LPMAP

Output a lineprinter map of R factors for all hkl and for each of 5 resolution shells.

NRFACTOR <nrfactor>

Number of R factors to be listed to output file. Default is 15; maximum is 500.

RESOLUTION [ <resmin> ] <resmax>

Low and high resolution limits in either order. These are in Angstroms or 4(sintheta/lambda)**2 if both are <1.0. Defaults: 1000.0 2.0

RESTART <nref> <ihh> <ikk> <ill>

Intermediate results will have already been written to a file assigned to SEARCHSAVE. Calculation is restarted after the <nref>-th reflection; <ihh> <ikk> <ill> are the indices of the first reflection to be included. These 4 numbers are printed to the log file after every 100 reflections.

RZONE <number>...

5 numbers to define a reflection zone limiting reflections included in the R value calculation.

To define the zone -h +k +l = 3n     input RZON    -1 1 1   3 0
To define                 l = 2n +1  input RZON     0 0 1   2 1

This keyword may be given more than once, but at present ALL conditions must be obeyed simultaneously.

SCALE <scale> <bfact>

The scale and temperature factor to be applied to observed structure factors FP. Defaults: 1.0 0.0.
FP(USED) is scale*FP(INPUT)*EXP(-bfact(SINTHETA/LAMBDA)**2).
It is sensible to run SCALEIT to put FP onto the same scale as one of the (partial) structure factors FCn before running RSEARCH. The scale here should then be approximately sqrt(nsymp), where nsymp is the number of primitive symmetry operators.

SYMMETRY [ <nspgrp> | <namspg> ]

Spacegroup number or name. You may alternatively input operators in X,Y,Z form. Spacegroup details are normally taken from the input MTZ file.

TITLE <title>

A title to be written to the log file.

GRID SPECIFICATIONS

XGRID | YGRID | ZGRID <min> <max> <n>

3 parameters for each of the 3 keywords: XGRID, YGRID and ZGRID.

The input is complicated to allow greater flexibility in describing the search volume. The first 2 parameters give the range in grid points, either directly or by reference to other limits given; the third, the number of grid points covering the unit cell in the appropriate direction.

   For a simple rectangular box :
    XGRID    ixmin  ixmax  nx    (3 numbers)
    YGRID    iymin  iymax  ny    (3 numbers)
    ZGRID    izmin  izmax  nz    (3 numbers)

   For a prism, there are many options. 
   For example :
    XGRID    ixmin  ixmax  nx    (3 numbers)
    YGRID   xmin  iymax  ny    (1 character string, 2 numbers) 
(or YGRID    iymin xmax  ny )  (number, character string, number)
(or YGRID   xmin xmax  ny )  (2 character strings, 1 number)
    ZGRID    izmin  izmax  nz    (3 numbers)

   where xmin means type the letters xmin ....

   For another sort of prism :
    XGRID    ixmin  ixmax  nx    (3 numbers)
    YGRID   xmin  iymax  ny    (1 character string, 2 numbers)
(or YGRID    iymin xmax  ny )  (number, character string, number)
(or YGRID     x    x   ny )  (2 character strings, 1 number)
    ZGRID    izmin  izmax  nz    (3 numbers)
(or ZGRID   xmin  izmax  nz )  (character string, 2 numbers)  
(or ZGRID    izmin xmax  nz )  (number, character string, number)
(or ZGRID     x    x   nz )  (2 character strings, 1 number)
(or ZGRID   ymin  izmax  nz )  (1 character string, 2 numbers)  
(or ZGRID    izmin ymax  nz )  (number, character string, number)
(or ZGRID     y    y   nz )  (2 character strings, 1 number)

Note that maps can only be output if the grid is rectangular. The maximum total grid points in the search volume is 60000. The maximum number in any one direction is 1024.

END

End of input.

INPUT AND OUTPUT FILES

 
INPUT  : MTZ Data file assigned to HKLIN
        [File containing intermediate sums assigned to SEARCHSAVE]
OUTPUT :[File for Binary R factor map assigned to MAPOUT]       

Those in parentheses are optional.

NOTES

Memory allocation

The program allocates memory dynamically. If the default isn't enough you can increase it by setting the logical name RSEARCH_NSIZE to the appropriate value indicated by the error message. (It ought to figure this out for itself.)

Suggested procedure leading to RSEARCH

Run LSQKAB to rotate your search coordinates by alpha beta gamma obtained from rotation search.
DO NOT FORGET to change your CRYSTL and SCALE cards to those of the new space group.
Remember that the rotation angles are relative to a given NCODE. The SCALE matrix MUST be consistent with NCODE.
COORDCONV will generate such a matrix when it converts fractional coordinates to PDB format.
Run SFALL to generate a list of h k l FC PC for all +-h +-k l for these coordinates in this cell.
Run CAD to collect together H K L FP SIGFP FC1 PC1 FC2 PC2 ... for the unique set of intensity data.
Run SCALEIT to scale FP to one of the FCn.

Run RSEARCH

#
# *****************************************************************
# 
#  Run sfall to calculate structure factors
#
sfall 
HKLOUT sfrkall.mtz 
XYZIN trans_pgk.pdb 
<< END-sfall                     
TITLE SFS
MODE  SFCALC XYZIN
GRID  80 160 160
RESOLUTION 20 3 
SYMMETRY 1
SFSGRP   1
LABOUT   FC=FC PHIC=AC
END
END-sfall
#
# *****************************************************************
# 
#  Run cad to collect FP and FC's into one file.
#  Up to this stage, can use the same procedure for TFFC & RSEARCH.
#  No phase changes made to partial phases.
#
cad 
HKLIN1 allderiv.mtz 
HKLIN2 sfrkall.mtz 
HKLOUT cad.mtz 
<< END-cad
RESOLUTION OVERALL 20 3
TITLE  resolving the bacillus pgk 
LABIN  FILE 1 E1=F  E2=SIGF
CTYPIN FILE 1 E1=F  E2=Q 
LABIN  FILE 2 E1=FC E2=AC
CTYPIN FILE 2 E1=U  E2=V
END
END-cad
#
# *****************************************************************
#  
#  Optional step 
#  Run scaleit to put FP onto the same scale as one of the FC's. 
#  Remember you will need to use the resolution required for RSEARCH.
#
scaleit 
HKLIN cad.mtz 
HKLOUT cad_sc.mtz 
<< END-scaleit
TITLE Scale FP to one FCpart ready for RSEARCH
RESOLUTION 8 4
REFINE   isotropic
LABIN    FP=FC1 SIGFP=FC1 FPH1=F SIGFPH1=SIGF
CONVERGE ABS 0.0001 TOLR 0.000000001 NCYC 4
END 
END-scaleit

EXAMPLES

For previous stages to this, see 'NOTES'

 
#  Rfactor search
#  Input mtz hkl list of FP SIGFP FC1 PC1 FC2 PC2.....
#  R factor requires FP scaled to FC. 
#  Correlation coefficient should be independent of scale.
#
rsearch 
HKLIN cad_sc.mtz 
MAPOUT dav.map 
SEARCHSAVE dav.tmp 
<< END-r  
TITLE Bacillus search in a-c plane 
SCALE FP 1.414 0
RESOLUTION  8 4           !  resolution limits
FOLIMITS 10 10000000000.  !  Fobs limits
FCLIMIT  50               !  Fc minimum
CORRELATION 
BINMAP                    !  binary map output
AXIS  Y Z X             
XGRID 0 20 40             !  grid limits 
YGRID 0 1  1 
ZGRID 0 40 80 
LABIN FP=F SIGFP=SIGF FC1=FC1 PC1=AC1 FC2=FC2 PC2=AC2  
END
END-r

AUTHOR

Eleanor Dodson, York University