*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
Hi Anthony and other CCP4ers,
Anthony Duff wrote:
*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
Its my experience that:
1. Simulated annealing, even without omitting anything, does a lot of
damage to a refined structure
also my experience - after the first hype with SA, I'm not using it for
many years anymore, for that reason.
2. CNS does a worse job of refining a structure in the late stages,
even accounting for differences in default restraint weights. (I
don't know why this would be so, with both using maximum likelihood...
maybe the CNS algorithms are inferior?)
Yes, the automatically determined "wa" weights in CNS are a factor 2-3
too small (at least in my experience). It is easy to modify the script
for automatic weight determination, accordingly.
My current opinion is that you shouldn't bother with SA-OMIT maps when
your structure is done, and that an ordinary refmac omit map is good
enough (set q=0 for omitted atoms, do enough rounds of refinement to
loose bias, then look at the ordinary DELFWT map). The rationale is
that in the finished structure, the phases are essentially correct and
you have got rid of superfluous bits of model (atoms not corresponding
with electron density) that are able to retain memory of the omitted
atoms.
Here, one has to be very cautious: it is generally safer to remove the
partial structures to be omitted from the coordinate file rather than to
set their occupancies to "0", even if the geometry at the boudaries will
get worse. The reason lies in the bulk solvent mask: if atoms with zero
occupancy are still considered in the bulk solvent mask calculation,
they leave a completely empty region in that bulk solvent mask. Since
there shouldn't be any empty region there (you will have either a more
or less ordered part of your structure or continuous bulk solvent
density), this empty region will show up as positive peaks in a
difference map (positve observed density minus empty density gives
positive peaks), thereby self-confirming what has been put in before.
This is exactly what you don't want in an omit-map! By removing the
atoms from the structure you avoid such holes in the bulk solvent mask.
I've seen exactly such self-confirming peaks in CNS difference maps by
setting the occupancies of omitted atoms to zero, instead of removing
them, and doing a mask bulk solvent correction. But maybe REFMAC5
ignores atoms with zero occupancy in the bulk solvent mask calculation
(Garib?)? If yes, it would be safe with REFMAC5 to set their occupancies
to zero, if no, better remove them from the coordinate file.
Best regards,
Dirk.
--
****************************************
Dirk Kostrewa
Paul Scherrer Institut
Life Sciences, OFLC/110
CH-5232 Villigen PSI, Switzerland
Phone: +41-56-310-4722
Fax: +41-56-310-5288
E-mail: [EMAIL PROTECTED]
http://sb.web.psi.ch
****************************************
