UB Home Page CCR Home Page HWI Home Page
DOWNLOAD
Announcements
DREAR
Bugs
FAQs
Funding
Manuals
Multiprocessing
Parameters
References
Personnel
Report Structures
SnB Server
SnB Successes
SnB Tutorial
Contact
SnB Home Page

Create |E|s

This screen provides the interface for Robert Blessing's DREAR suite of data-reduction programs (Data Reduction and Error Analysis Routines). These programs are used to compute the normalized structure-factor magnitudes (|E|s) required for direct-methods phasing.

  • Native & Derivative Input File Names: Enter the name(s) of your reflection data file(s). Basic and SAS applications require only native data. SIR applications require input reflection files containing the native and derivative data. Supply the path designation if the file is not in the directory where the SnB GUI is running.
     
  • Native & Derivative Input File Types: Select the input reflection file format from the following choices:
     
    • An free-format ASCII file consisting of records containing either H, K, L, F, Sig(F)
      or H, K, L, F^2, Sig(F^2) with fields delimited by blanks. The program "mtz2various"
      can be used to convert CCP4 reflection files into this format.
       
    • SCALEPACK (unique anomalous data) output files:
      The proper file has three header records (ignored by SnB) followed by
      reflection records H, K, L, I+, Sig(I+), I-, Sig(I-), containing the anomalous
      dispersion information (when present) on a single line.
       
    • d*TREK output files:
      1. One record containing 3 integers (N_INT + N_FLOAT + N_STRING) where
          N_FIELD = N_INT + N_FLOAT + N_STRING
      2. N_FIELD records ignored by SnB
      3. Free-formatted reflection records containing H, K, L, I, Sig(I) or
           H, K, L, (N_INT-3) integers, I, Sig(I). For SAS applications, Bijvoet-related
           measurements must be present as separate records.
       
  • Output File Name: Enter the name for the reflection file containing |E|s that is to be output after normalization by the DREAR routines. A suggested entry is "structure_name.drear".
     
  • Native & Derivative ASU Contents: For substructure applications, the complete native ASU elemental contents must be specified for both SIR and SAS data sets; derivative ASU contents are required for SIR data only. The anomalously scattering element MUST be included for SAS data. Exact atom counts are not required for the lighter elements, and C, N, and O can be combined if desired. It is not necessary to include solvent, but a reasonable approximation for the solvent content certainly would not hurt. The following approximation is suggested for a protein with R residues: C 5R, N 1.2R, O1.5R, and H 8R. Remember that Se replaces S in methionine residues only.
     
  • Data Resolution Range: Only reflections within the specified resolution limits will be normalized and included in the output file. During an initial run, it is suggested that all data be included and the output of programs EVAL (eval.lp) or DIFFE (diffe.lp) be inspected to determine whether the data scale well over all resolution ranges. The average values of |E|^2 should equal unity for all resolution ranges. If the observed averages are less than 0.5 or greater than 2.0 at either resolution extreme, it may be useful to apply appropriate limits and rerun the normalization procedure (i.e. rerun DREAR). Optionally, resolution cut-offs are available from the reflections & invariants screen, and use of these cut-offs does not require rerunning DREAR.
     
  • Use Bayesian estimates?: The Bayesian correction for weak data (such as negative intensities) should be applied if a similar correction has not already been made. Thus, "No" should usually be selected if the input data are in the form of |F|s. This correction should NOT be applied to Fa values.
     
  • Use locally normalized |E| values?: The use of |E|s that have been locally normalized by program BAYES may sometimes be helpful for low-resolution data (lower than 3.5A). This procedure is frequently described as normalization in resolution shells. Indicate "Yes" to invoke this option. The more common procedure is global normalization via the Wilson plot (programs LEVY and EVAL), and this is the default option for higher-resolution situations.
     
  • Perform local scaling for difference data?: Local scaling [Matthews, B.W. & Czerwinski, E.W. (1975). Acta Cryst. A31, 480-497.] may be advisable for SIR and SAS data, but it has not been proven conclusively to increase the success rate for SnB applications. Local scaling is provided here as an option, but the default is not to use it. The use of local scaling is an option that may be considered if initial applications do not lead to a solution.
     
  • Options for difference |E| (diffE) computation - SIR & SAS data types: Successful phasing of difference magnitudes requires careful screening and elimination of erroneous measurements on account of the inherently low signal-to-noise ratio for such data. The input cut-off parameters (Tmax, Xmin, and Ymin) and the output cut-off parameters (Zmax and Zmin) for the diffE program have been defined in the following reference: Blessing, R.H. & Smith, G.D. (1999). J. Appl. Cryst. 32, 664-670. A detailed investigation of these parameters for S-adenosylhomocysteine hydrolase has led to the suggested default values and is described in the following: Howell, P.L., Blessing, R.H., Smith, G.D. & Weeks, C.M. (2000). Acta Cryst. D56, 604-617. Zmin is the last cut-off to be applied, and it's value is specified on the next screen (reflections & invariants). This permits the user to vary the final cut-off without rerunning the entire DREAR package.
     
  • Executing DREAR Programs: In order to execute the DREAR suite of programs, simply press the Execute DREAR Suite button. Once inspection of the DREAR results has been completed, it is advisable to press the Clean DREAR Files button to remove the many intermediate output files that get created. Clicking on View DREAR Results will display the top of the final output reflection file. The pop-up window displays the largest |E|s sorted in decreasing order, and it should be inspected carefully for the presence of unreasonably large values or the occurrence of a non-random distribution of Miller indices among the largest reflections. The presence of either condition is an indication of possible problems with the data. The presence of many reflections with low |E|/ sig|E| ratios at the top of the list may also indicate that there will be too few remaining reflections after the Zmin cut-off is applied.