*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
Hi Xiongying,
Using the mainchain dihedrals as a real-space "free R-factor" is one
reason not to restrain them, but if there was valuable information in
the prior knowledge about prefered dihedrals then it would be a waste to
ignore it. At least you could impose the restraints in the final
refinement run, just like you should use the Rfree set reflections in
your final refinement run (using the same refinement parameters that
were shown to work well when you did not use the Rfree reflections).
In past trials I never found restraints of mainchain dihedrals to be of
much help, most likely because they don't have much information content.
The rmsd of phi/psi angles in the database for a particular, let's say
helical, conformation is relatively large and, accordingly, the
restraint you can impose is weak. Basically, the information content of
the diffraction data dominates the dihedral restraints so most of the
time imposing them is a waste of your time.
The one situation that led me to explore this was when I was using Xplor
with its simulated annealing torsion angle refinement. In my hands,
Xplor introduced as many errors as it fixed (it especially messed up
leucine residues that have two conformations that fit the density well
but only one has a favorable dihedrals). So my thought was that if I
could put a weak unimodal restraint on all dihedrals I was confident
about then XPLOR could only work and, hopefully, improve the trouble
spots. It never quite worked out for me but the idea may still have
merit as I didn't try very hard.
Bart
Gerard DVD Kleywegt wrote:
*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
Does anybody know how to include phi/psi dihedral restraints to
minimization or torsion annealing in CNS? Many Thanks
you have to ask yourself if you really, really, really want to do
that. and if the answer is 'yes' you have to ask yourself again
the reason is that the conformational torsion angles are eminently
suited for validation purposes, but only when they are unrestrained.
in other words, once you restrain phi and psi, you can't use
ramachandran plots any longer (well, you can, but they are useless as
validation tools)
if you still want to do it, there are two ways i can think of. one is
to use the database-derived torsion restraints of kuszewski and clore
- these used to be in xplor, not sure about cns. however, this is not
recommended until the model is nearly finished because these
restraints tend to pull your residues into the *nearest* well of the
'ramachandran energy' which is not necessarily the same as the
*correct* well
the other method is to restrain the torsions in your new crappy 3 A
model to the values they had in your bright and shining 1.5 A model.
you can use XPLO2D to generate the restraints file which you can
include into your xplor script (and i reckon it should work with cns
as well). for a description, see:
http://xray.bmc.uu.se/usf/xplo2d_man.html#H5
--dvd
******************************************************************
Gerard J. Kleywegt
[Research Fellow of the Royal Swedish Academy of Sciences]
Dept. of Cell & Molecular Biology University of Uppsala
Biomedical Centre Box 596
SE-751 24 Uppsala SWEDEN
http://xray.bmc.uu.se/gerard/ mailto:[EMAIL PROTECTED]
******************************************************************
The opinions in this message are fictional. Any similarity
to actual opinions, living or dead, is purely coincidental.
******************************************************************
