Dear Gerard and Ian,
thank you for your work in putting together the Staraniso sever and your
effort to tackle the anisotropic diffraction problem, a non-trivial
behavior that creates many problems in structure solving, and that is
very prevalent in the world of membrane proteins...
I have been following your threads very closely over the past months; I
am now in the possession of another one of these anisotropic datasets
and just submitted a job on your server, and I would like to ask you
some general questions for which I couldn't find answers on your website
(I'm emailing the ccp4bb as it might be of interest to other readers as
well).
* It seems like the (local) I/s(I) is your criteria of choice to decide
how far is the data diffracting. And you offer us a range of values from
1 to 3 with small steps, and a default value of 1.20. Could you shed
more light on this choice, how important it is, and what attention
should be devoted to it?
My understanding from this is that with the new detectors, a photon
measured is a real photon, so users can go to very low values of I/s(I)
to get the most data out the detectors. For isotropic diffraction, I
would cut my data at several of these values and consider the maps to
see if I have any gain on their quality. However, with my anisotropic
data, with the default cut off of 1.20, I get in the following table
that brings me more questions:
Summary of merging statistics for observed data extracted from the final
MRFANA log file:
Overall InnerShell
OuterShell
---------------------------------------------------------------------------
Low resolution limit 48.618 48.618 4.200
High resolution limit 3.811 12.254 3.811
Rmerge (all I+ & I-) 0.182 0.046 7.878
Rmerge (within I+/I-) 0.186 0.043 7.643
Rmeas (within I+/I-) 0.223 0.053 9.089
Rmeas (all I+ & I-) 0.200 0.052 8.537
Rpim (within I+/I-) 0.122 0.031 4.870
Rpim (all I+ & I-) 0.080 0.024 3.262
Total number of observations 148030 5479 8188
Total number unique 23914 1192 1196
Mean(I)/sd(I) 7.3 21.2 1.3
Completeness (spherical) 60.4 96.9 12.0
Completeness (ellipsoidal) 92.5 96.9 67.0
Multiplicity 6.2 4.6 6.8
CC(1/2) 0.996 0.996 0.061
* In the past, before your server existed, I have cut my anisotropic
data keeping a 20% completeness in the highest resolution shell, it
seemed to me the correct cutoff to ensure map quality, albeit I must say
it was more a "wet-finger" guess than a real study. This completeness
was of course spherical. Given the I/s(I) criteria stated above, is the
completeness a secondary criteria or should it be taken into account to
refine which I/s(I) to choose for scaling performed in Staraniso? What
completeness should be kept minimal?
* A side question arises from this table, I then have a CC(1/2) of 0.061
in the highest resolution shell. Should it be worrisome or is CC(1/2)
not a good criteria in this case? Same question as above, what CC(1/2)
should be kept minimal?
Thank you in advance for your help and feedback.
All the Best
Vincent
--
Vincent Chaptal, PhD
Institut de Biologie et Chimie des Protéines
Drug Resistance and Membrane Proteins Laboratory
7 passage du Vercors
69007 LYON
FRANCE
+33 4 37 65 29 01
http://www.ibcp.fr