dtprocess on-line help

Scale/Average: Scale and average a reflection list into unique hkl,Intensity,SigmaI

The Scale/Average menu prepares a dtscalemerge command and runs it. Since dtintegrate does the equivalent of what is sometimes called postrefinement, there is no need for postrefinement by dtscalemerge.

Select a header from the top list. Usually this is dtrefine.head or dtintegrate.head.
Select the reflection list to scale and average from the reflnlist lists.
Set the menu options described below.
Select Run scale
View the dtscalemerge log file.

Scale/Average menu options

Exclude sigma: Input reflections with I/sigmaI less than the Exclude sigma are excluded from contributing to the scale factor refinement. However, these reflections are included in the final statistics.
Rejection sigma: Sets the rejection level for reflections. Reflections with scaled intensities that differ by more than the Rejection sigma from the weighted average intensity calculated by other symmetry-related reflections are flagged as rejected by setting their observed standard deviations to be negative. Default: 1x10e11, so no reflections are rejected.
Weight multiplier
Weight addend: These factors alter the error model and weights applied to the observed reflections according to Whj = 1 / [ (Sighj * Weight multiplier)^2 + (Ihj * Weight addend)^2]
Resolution: Only reflections within the resolution limits in Angstroms are used scaled, averaged and written to the output files. If either limit is 0.0, then resolution limits are not imposed.
Fixed batch: Specifies the name of the batch whose scale factors will remain fixed. The scale factors of all other batches will shift relative to this batch. Default: first batch in the input reflection list.
Initial fixed scale: Specifies the scale factor kj of the fixed batch. Default: 1.
Max cycles: The maximum number of non-linear least squares cycles to perform. If the refinement converges, then the maximum number of cycles may not be reached. No shifts are applied the last cycle.
Initial fixed B: The B-factor Bj of the fixed batch.
Refine B-factors: Refines or fixes the so-called B-factors which are all set to 0.
Output anomalous I+,I- on SAME line: If set, the output scaled and averaged reflection list will have not only the merged intensity from all observations (all I+ and I-) in the fIntensity field, but additional fields for separately merged I+ and I- observations. The fields for each merged reflection are thus nH, nK, nL, fIntensity, fSigmaI, fIntensity+, fSigmaI+, fIntensity-, fSigmaI-. Since a centric measurement can be assigned to an I+ or I-, even merged centric reflections may have different values in the fIntensity+ and fIntensity- fields. Missing information is indicated by a -1 value.
Scale anomalous, I+,I- separately: If set, I+ and I- are considered unequal and kept separate. This means that Rmerge, rejections and averaging all keep I+ and I- separate. The output file will have I+ and I- on separate lines. This turns OFF the -anom option (Output anomalous I+,I- on SAME line). This is useful if you have a strong anomalous signal and high redundancy.
Output scaled, unaveraged reflns: If set, the scaled but unaveraged reflections are written to the specified file. This may be useful for further scaling or for those that wish to run their own averaging and rejection algorithms. Reflections rejected by dtscalemerge will have their SigmaI values set to 0 or less.
Output hkl,I,sigI ... noheader: If set, the scaled, but unaveraged reflection list file will have only h, k, l, Intensity and sigmaI fields in free format and no header. Rejected measurements (sigmaI <=0) will be excluded from this file. This format is suitable for input to many non-d*TREK programs such as teXsan for subsequent processing.
Output scaled, averaged reflns: The name of the output reflection list file. The name cannot begin with - character. If blank: Do not write an output scaled and averaged reflection list file. The file is in d*TREK format with fields nH, nK, nL, fIntensity, fSigmaI.
Output ... noheader: If set, the scaled and averaged reflection list file will be written without a header. This is sometimes useful for subsequent processing with non-d*TREK programs.