*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
For PDB and publications you should state statistics for the data that
you actually used for structure determination. Clear and simple.
Deciding what data to use and what data to cut is less clear and this
has been discussed on this list before and is currently also a topic of
discussion at Acta Cryst. Basically you keep all data from the lowest
resolution to wherever you feel there is still real information that
contributes to the quality of the map & model. Some have argued to cut
at a resolution where I/SigI drops below two, others use 1.5 and yet
others think I/SigI can be manipulated so much by changing the error
model that it is not a clearcut situation. What you definately don't
want to do is to reject weak data within your selected resolution range.
Weak data may not contribute much to maps but they are perfectly fine
observations otherwise.
Bart
Florian Schmitzberger wrote:
Dear All,
I would have a question about reporting diffraction data quality
indicators (for pdb-submission and publication) for data processed with
XDS/XSCALE. XSCALE(.LP) lists data quality indicators such as
completeness, mean <I>/sigma<I>, R-meas etc. for several signal/noise
cutoffs descending to negative values for the latter (up to -3.0).
I was wondering whether it makes more sense to report data quality for
data with a signal/noise > 0 (or indeed 1) rather than report the data
quality for (all) data with a signal/noise >-3.0.
Whereas I imagine for error estimates and standard deviation
calculations (in ML-based refinement) including the quality of the
negative intensity reflections (signal/noise >-3.0) is still meaningful
to know, these reflections probably did not contribute significantly to
the diffraction, calculation of the electron density, and the final model.
I am asking this since I observe the following difference (I suppose
because of the altered sd estimates) between the tables for the 0.0
(Rmeas=94% (flame on, I am aware) and I/sigmaI= 1.83) and -3.0 (and
Rmeas=106.4%, I/sigmaI= 1.51) signal/noise cutoff.
I am aware this question is of a rather cosmetic nature and the data
will of course be the same (but reporting may have slightly different
effects on referees, I could imagine).
Thank you very much in advance for any comments!
Regards,
Florian
--------------------------------------------
Florian Schmitzberger
Medical Biochemistry and Biophysics
Karolinska Institute
Scheeles vaeg 2
SE-171 77 Stockholm, Sweden
Tel: +46-8-524-86875
--
==============================================================================
Bart Hazes (Assistant Professor)
Dept. of Medical Microbiology & Immunology
University of Alberta
1-15 Medical Sciences Building
Edmonton, Alberta
Canada, T6G 2H7
phone: 1-780-492-0042
fax: 1-780-492-7521
==============================================================================
