Preparation of diffraction data

We will assume that the diffraction data has been converted to CNS format using to_cns or CCP4. There will then be a reflection file containing the observed structure factor amplitudes and estimated errors (sigma values). Cross validation will be used to monitor the subsequent refinement of the model and also to reduce bias in map calculations. The cross validation information is added to the reflection file using the CNS task file make_cv.inp.

      cns_solve < make_cv.inp > make_cv.out [6 seconds]

A new reflection array is created, by default the name TEST is used. The array contains integer numbers, by default 1 signifies that the reflection will be used for cross-validation (i.e. excluded from refinement). The output of the make_cv.inp task file is a CNS formatted reflection file. This will contain a header which describes the data items in the file, followed by the data items themselves. The file is plain ASCII and can be viewed with any text viewer/editor. Example header from a CNS reflection file with cross validation information:

 NREFlection=     25615
 ANOMalous=FALSe { equiv. to HERMitian=TRUE}
 DECLare NAME=FOBS                   DOMAin=RECIprocal   TYPE=REAL END
 DECLare NAME=SIGMA                  DOMAin=RECIprocal   TYPE=REAL END
 DECLare NAME=TEST                   DOMAin=RECIprocal   TYPE=INTE END
 INDE     0    0    3 FOBS=  1047.380 SIGMA=    18.630 TEST=         0
 INDE     0    0    6 FOBS=   197.360 SIGMA=     2.980 TEST=         0
 INDE     0    0    7 FOBS=   451.840 SIGMA=     4.920 TEST=         0
 INDE     0    0    8 FOBS=   410.340 SIGMA=     4.270 TEST=         1
 INDE     0    0    9 FOBS=   666.420 SIGMA=     8.190 TEST=         0

The phase information to be used in the refinement will be the Hendrickson-Lattman coefficients obtained from MAD phasing (see other tutorials). It is sufficient to just add the output .hkl file, from the final MAD phasing job, to the list of reflection files read into the CNS task files. It is important to list this file last if the other reflections files do not contain anomalous data.

In some cases the experimental phases will have come from a different program and therefore will be in a different format. If the format is plain text then an awk or perl script can usually be written to convert to CNS format. For binary files such as the CCP4 MTZ format a special program will be required to convert to CNS format. There are 2 such programs for this (both of which are part of CCP4 version 3.5):

• SFTOOLS: This can read an MTZ file then write it back out in CNS format. All items will be correctly converted. However, the header information from the CNS file is not written. This must be added manually. The header for a file with experimental phase information should look something like this:

   NREFlection=           1400
   ANOMalous=TRUE { equiv. to HERMitian=FALSe}
   DECLare NAME=FOBS        DOMAin=RECIprocal   TYPE=COMP END
   DECLare NAME=SIGMA       DOMAin=RECIprocal   TYPE=REAL END
   DECLare NAME=PA          DOMAin=RECIprocal   TYPE=REAL END
   DECLare NAME=PB          DOMAin=RECIprocal   TYPE=REAL END
   DECLare NAME=PC          DOMAin=RECIprocal   TYPE=REAL END
   DECLare NAME=PD          DOMAin=RECIprocal   TYPE=REAL END
   GROUp TYPE=HL
     OBJEct=PA
     OBJEct=PB
     OBJEct=PC
     OBJEct=PD
   END
   INDE     0    4    2 FOBS=     677.8       0.0 SIGMA=     19.40
                        PA=    -0.486 PB=     0.000 PC=     0.000 PD=     0.000
  

  The grouping for the HL coefficients is essential in order that remapping of reflections to a different asymmetric unit is performed correctly.
• mtz2various: This can read an MTZ file and write out a CNS reflection file. An example script is:

  mtz2various HKLIN infile HKLOUT outfile << EOF
  RESOLUTION 10000 2
  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
  

In some cases the experimental phase information is only available in the form of centroid phases and figures-of-merit (FOM). These must be converted to Hendrickson-Lattman coefficients in order to be used in refinement in CNS. The CNS task file make_hlcoeff.inp can be used for this purpose.

In general it is best to use the "raw" experimental phase information, i.e. no density modification or other phase improvement methods have been used. Density modified phases usually have inflated FOM estimates which will lead to their over-weighting with the MLHL target in refinement. Therefore, if density modified phases are to be used it may be wise to "blur" the probability distributions by the application of a scale factor and B-factor. The CNS task file hlcoeff_blur.inp can be used for this purpose.

It is a good idea to check that the electron density map generated from these experimental phases after conversion to CNS format is unchanged and that the model coincides with the density. The CNS task file fourier_map.inp can be used to make a Fourier synthesis using the observed amplitudes and the experimental Hendrickson-Lattman coefficients.