I find this confusing also!
Phil Jeffrey wrote:
On 1/13/11 2:48 PM, J. Fleming wrote:
Hi All,
I'm about ready to deposit my structure and have used pdb_extract to
aid in the process. Unfortunately the following values were not found
and are required by ADIT:
1) Under Data Collection, Reflections section: Observed criterion
sigma(F) and Observed criterion sigma(I)
There is no criterion for sigma F applied in Denzo/HKL2000. Not least
of all because data processing programs like Denzo and Scalepack work
with intensities and not structure factor moduli.
Both are marked as required, but if you leave one blank the deposition
will go through anyway and the annotator will not have a problem with this.
Just as "synthetic source" and "engineered source" are required, but you
leave them blank or NULL if your protein is from a natural source.
The default Sigma(I) cutoff is -3
See:
http://www.hkl-xray.com/hkl_web1/hkl/Scalepack_Keywords.html (keyword
SIGMA CUTOFF)
This is more complicated if there are partials -
The -3 Sigma cut-off applies to "measurements"
"measurements" can be full or partial (I think).
I believe the reflection is rejected if any of its partials are rejected-
(although theoretically it should be possible to approximate an intensity
from intensity of one of its partials)
The number rejected will be greater if only one of the partials needs to
fail the sigma cutoff than if they are all added up before the cutoff is
applied
2) Under Refinement, Refinement Statistics section: Number unique
reflections (all)
I think in refinement this just means test + working reflections?
In data reduction it is more complicated.
I think you have the following:
measurements - number of spots for which denzo recorded a complete reflection
or a full set of partials (different from "measurements" above?)
reflections, possible - Theoretical number unique reflections in range
reflections, measured - Unique reflections with at least one measurement
reflections, observed - whats left, after aplying rejection criterion
during data reduction. I would say once a reflection has been measured,
it has already passed the -3sigma cutoff, we never reject a measured
reflection.
reflections used in refinement - after applying resolution and sigma cutoffs
in refinement
(Don't take my word for this- I'm confused, as you can see. Hoping for
clarification)
Another question- it is often suggested to use I/sigI = 2 in a shell as
the criterion for resolution cutoff. And I/sigI is one of the parameters
needed for deposition. But is this the average I/sigI for individual
measurements, which (I think) is what you can estimate from the last table of
scalepack output, or the average of I/SigI for the reduced unique reflections,
which you could get from a table in truncate log file?
Sigma I for the reduced reflection should be smaller than sigmaI for
the individual measurements, as standard error of the mean is smaller
than standard deviation of the measurement.
