*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
On Jul 24, 2006, at 1:15 PM, Mark A. White wrote:
Bill,
At 2.0 A your bond rmsd should be in the 0.018-0.022 A range.
I know 0.005 is excessively conservative (the library probably has an
error close to double that) but it illustrates the point that freeR
doesn't improve when I weight the X-ray term more heavily. When I
really crank it up, I can get to 0.023 A (see below) but Rfree
becomes slightly worse (Rwork gets lower):
Just to add one more data point:
Rwork = 0.145 % Rfree = 23.1 %, rmsd bond = 0.023 A
Rwork = 15.2 %, Rfree = 22.6 %, rmsd bond = 0.010 A
Rwork = 16.7 %, Rfree = 22.6 %, rmsd bond = 0.005 A
However, this is for proteins and RNA is a different animal.
Apart from being vastly more interesting, it behaves rather similarly
(at least in CNS and refmac).
My feeling is that
RNA and DNA stereochemistry parameters are not as well defined as
those
for proteins. Illegal puckers and conformations are permitted by some
refinement programs, so increasing the x-ray weight will not improve
your model if the basic conformation is incorrect.
Refmac doesn't restrain the pucker like CNS does, but this actually
is better. I have had CNS move atoms out of the density to conform
to sugar pucker it wrongly thinks is required.
A recent article in
Acta Cryst. (which I cannot find at this moment) suggested that
many of
the RNA/DNA structures in the PDB are incorrect for this reason.
They probably looked at only mine. ;)
So what do I suggest? Do not let the x-ray weight term become too
small, increasing the weight to a reasonable extent, if it does not
improve Rfree still should not hurt your model (I expect a flame on
this
one).
Not unless you cc the bb.
Remember that adding waters adds 4 free parameters per water.
This happens before I add in the waters, i.e., it is the Mn++
These parameters can absorb the errors and signals which would cause
your bond lengths to vary. So that the more waters you add, the lower
your bond rmsd should be.
The Mn2+ ions are the strongest scatterers in the crystal. They are
also
unbonded, meaning that they can easily move around. You could try
adding ideal bond restrains to the Mn2+ ions to prevent them from
moving
too freely. This should lower your Rfree.
They are currently restrained to 2.0 A for phosphate oxygen chelation
and 2.4 A for N1 purine chelation. I think the fact that they are
the strongest scatterers, and that this data was collected on the
inflection (maxing the abs. value of the dispersion term), might be
the culprit. That's about 1.8 A wavelength. Our original intent was
to simply identify Mn++ sites, but the structure, as always, turned
out to be more interesting than I thought it would be. (Crystals with
only RNA and water molecules refined to 2.2 A in refmac have an Rfree
of about 23% and an Rwork of a little better than 20% with rmsd bond
0.009 A. These data were near the Br edge, around 0.9 A wavelength.
So maybe phosphori absorption and the large dispersion term at 1.8 A
are hosing us.)
(Phil Evans hopefully remembers the RNA credo: expect nothing, and
you can only be pleasantly surprised.)
Thanks very much for the advice. You've given me lots to consider.
All the best,
Bill
Good Luck,
Mark
On Mon, 2006-07-24 at 09:27 -0700, William Scott wrote:
*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
Hi folks:
I am refining an RNA into which Mn++ was soaked. The crystals
diffract well to 2.0 A resolution. I've identified the Mn sites using
anomalous differences and the sites are very clear and make good
chemical sense. I've also added ca. 200 waters using COOT to add
them, to do sanity checks, etc, and have also manually purged
anything that doesn't look sensible. In any case, the following
arises in the absence of the waters as well as their presence:
I have two identical refinements with the following statistics:
Rwork = 0.145 % Rfree = 23.1 rmsd bond = 0.023 A
Rwork = 15.2 %, Rfree = 22.6 %, rmsd bond = 0.010 A
Rwork = 16.7 %, Rfree = 22.6 %, rmsd bond = 0.005 A
If I do idealized refinement, I can't get the geometry much tighter
than 0.005 A. So it seems any relaxation of the constraints does
nothing for the free R. In addition, The gap between Rwork and Rfree
is significantly larger than in the refinement of the same crystals
in the absence of Mn.
The maps, etc. look sensible. I've done a composite omit map on the
whole structure (in CNS) as a sanity check and the RNA is definitely
correct.
I have to conclude I am doing something stupid. I would be grateful
for any suggestions.
Bill
Sincerely yours,
Mark A. White, Ph.D.
Assistant Professor, Dept. Biochemistry and Molecular Biology,
Manager, Sealy Center for Structural Biology and Molecular Biophysics
X-ray Crystallography Laboratory,
Basic Science Building, Room 6.660 C
University of Texas Medical Branch
Galveston, TX 77555-0647
Tel. (409) 747-4747
Fax. (409) 747-4745
mailto://[EMAIL PROTECTED]
http://xray.utmb.edu
http://xray.utmb.edu/~white
