Content:


Current defaults

In the current autoPROC version, the (final) criteria to reach in the outer (highest resolution) shell for different types of statistics are

Statistic Criterion
I/sigI >=2.0
R(pim) <=0.6
CC(1/2) >=0.3

In order to handle situations where a much too optimistic amount of data is given to the scaling procedure (e.g. the detector distance was set too close or the crystal diffracted worse than anticipated), those final values are reached iteratively, starting from more loose criteria.


Set your own values

To modify the default behaviour, the following parameters (a space-separated list of value1:value2 pairs) need to be modified - either on the command-line or in a so-called macro file:

  • Each pair defines a criterion for individual runs and one for overall datasets:
    • This is important for multi-sweep datasets where several separate sweeps (runs) will be combined/merged into a single dataset.
    • Higher multiplicity should result in better statistics - or to look at it the other way: lower multiplicity will give worse or unreliable statistics, which is why we don't want to be too restrictive with those individual runs. This is also the reason why the defaults (and the examples below) usually use slightly lower quality criteria for individual runs/sweeps compared to overall datasets.
    • For single-sweep datasets, the second value (dataset-specific) is the relevant one.
  • Within the iterative scaling procedure, each cycle N will take the Nth value pair - or the last one if fewer than N are defined)
Criterion Parameter Default Example Remark
I/sigI ScaleAnaISigmaCut_123 "0.1:0.1 0.5:0.5 0.5:1.0 1.0:2.0" "0.0:0.0 0.2:0.2 0.4:0.5 0.4:1.0" the final cut-off criterion is 1.0 (or 0.4 for each run if the final data is obtained by merging several sweeps)
R(pim) ScaleAnaRpimallCut_123 "99.9999:99.9999 0.9:0.9 0.8:0.8 0.6:0.6" "99.9:99.9" ignore Rpim as cut-off criteria
CC(1/2) ScaleAnaCChalfCut_123 "-1.0:-1.0 0.0:0.0 0.1:0.1 0.3:0.3" "-1.0:-1.0 0.0:0.0 0.1:0.1" use CC(1/2) of 0.1 as final criterion (for individual runs and overall dataset)

If one runs autoPROC or aP_scale e.g. with

% process \
  ScaleAnaISigmaCut_123="0.0:0.0" \
  ScaleAnaRpimallCut_123="99.9:99.9" \
  ScaleAnaCChalfCut_123="0.3:0.3" ...
% aP_scale \
  ScaleAnaISigmaCut_123="0.0:0.0" \
  ScaleAnaRpimallCut_123="99.9:99.9" \
  ScaleAnaCChalfCut_123="0.3:0.3" ...

the only two active criteria (right from the first cycle of iterative scaling) would be I/sigI (at least 0.0 in the highest resolution shell) and CC1/2 (at least 0.3). For the above reasons it might be better to use a multi-stage definition:

% process \
  ScaleAnaISigmaCut_123="0.0:0.0" \
  ScaleAnaRpimallCut_123="99.9:99.9" \
  ScaleAnaCChalfCut_123="0.1:0.1 0.2:0.2 0.3:0.3" ...
% aP_scale \
  ScaleAnaISigmaCut_123="0.0:0.0" \
  ScaleAnaRpimallCut_123="99.9:99.9" \
  ScaleAnaCChalfCut_123="0.1:0.1 0.2:0.2 0.3:0.3" ...

STARANISO (anisotropic analysis)

Remember that the analysis for anisotropy via STARANISO uses a local I/sigI criterion, independent of the above system. It uses a default value of 1.2 for a particular measurement to be classified as being observed.


Typical examples

To use only CC(1/2) as criterion (with the default of 0.3), just run with

% process -M HighResCutOnCChalf ...

To adjust the I/sigI criteria downwards to a (final) value of 1.0 - but still keeping the CC(1/2) and R(pim) criteria - use e.g.

% process ScaleAnaISigmaCut_123="0.1:0.1 0.5:0.5 0.5:1.0"

If no automatic decision regarding the high-resolution limit should be done at all, running

% process -M NoHighResCut ...

or (to use own isotropic resolution limit)

% process -M NoHighResCut -R 50.0 2.8 ...

should have the desired effect. An initial autoPROC run without those explicit resolution limits would be advisable - especially for careful analysis of the STARANISO results that could otherwise be invalidated by a too strict initial resolution limit. However, if only a small, central region of the detector surface contains diffraction a generous, but more sensible high resolution limit could be imposed frm the start in this way.