Clive Delmonte wrote:
> Am I right to think that, were the smaller unit cell dimensions to be used
> in the deconvolution of the structure factor files, perhaps re-indexing as
> necessary, and whether using MAD, heavy atoms, etc., the resulting
> Pattersons would merely emerge with smaller scales on their X,Y axes ?
> Could not the phasing process and subsequent unbiased refinement be carried
> out equally well on the smaller unit cell, but inevitably result in a
> different structure ?
The unit cell determined from an X-ray crystal is actually pretty well
determined (assuming the wavelength has been measured correctly and
the detector settings are correct - these are usually checked by
measuring the scattering from a standard wax sample).
However, if the cell were measured incorrectly, the structure solution
by heavy atom/MAD methods could continue as normal. The features of the
Patterson would simply be rescaled. The scale and temperature factors
of the heavy atoms would be systematically altered, but the positions
(in fractional coordinates) would be unaffected.
The heavy phasing calculation should also procede succesfully, and so
the first electron density map will show the same features.
However, attempts to build into this map will cause problems, because
the size and spacing of the features will be wrong. Whilst the shape
of the electron density will be unequivolcal, its interpretation in
terms of atomic positions will have to change since the bonding
distances must remain resonable. At lower resolutions it might be
possible to build an incorrect model into the electron density, but
this should be detected by the cross-validation test. Restrained
refinement is likely to fail catastrophically.
At high resolutions where the individual atomic peaks are resolved,
the error in the cell dimensions wiill become immediately obvious
from the distribution of inter-peak distances. Such maps are more
likely to be obtained using direct-methods phasing rather than
MAD/MIR, but I think a few MAD structures have been done at very
high resolutions.
Kevin
--
Dr. Kevin Cowtan, Structural Biology Laboratory, University of York, UK
