On Fri, 2007-01-26 at 08:23, Paul Emsley wrote:
> Yes, sorry about that. Fixed in pre-release :-/
> For now, to unset them you need to click on the "No Sec Str Restraints"
> radiobutton.
>
> Richard Baxter and I discussed the specific problems in private. I
> hope that he will summarize soon.
My apologies for not summarizing results of my previous query:
Sue Roberts provided the following script commands sent by Kevin after
her own query:
(add-planar-peptide-restraints) this is also in the manual
(set-matrix 20) default 60, lower means stronger
stereochemistry weight
(set-terminal-residue-do-rigid-body-refine 0)
(set-add-terminal-residue-do-post-refine 1)
(set-mutate-auto-fit-do-post-refine 1)
These switches control placing of residues from the "Add terminal
residue" command.
Real-Space Refinement can be expected to produce distortions give low
resolution maps.
I sent Paul an example where the backbone appears broken after
regularization, but this is in fact a case where the previous placement
of the residues allows no "good" sterochemistry. Regularize increased
the CA-C bond distance to 1.7 Angstrom making the bond appear broken but
it is not, if you Rotate/Translate the carbonyl moves with the correct
residue and after manual adjustment and further regularization it is
fine.
One thing I have noticed that may confuse Regularize is the position of
the carbonyl oxygen relative to the peptide bond. After manually
addition it may often be on the wrong side, and this is not fixed by
editing backbone torsions because that also moves the C atom and messes
up stereochemistry at the CA. In such cases I find it best to manually
move the carbonyl oxygen prior to Regularize, select Rotate/Translate
and then [Cntrl]+left-mouse to drag a single atom.
> And just to be double clear, we are talking about *Regularize Zone*
> here, aren't we?
> You should be getting Chi-squareds less than 0.001. Are you?
> Coot tells you in the console if it is applying Sec Struct Restraints
> to your zone.
>
> Paul.
>
>
> On 26 Jan 2007, at 13:55, Joel Bard wrote:
>
> > Hi-
> >
> > One thing I've noticed is that if you turn on secondary structure
> > restraints in the Refine/Regularize Control GUI and then click OK,
> > these restraints stay on even though subsequent clicking of the
> > Refine/Regularize Control show the No Secondary Structure Restraints
> > radio button selected. To unset the secondary structure restraints I
> > click Alpha Helix Restraints and then click No Secondary Structure
> > Restraints and then hit OK (though I haven't tested that this is
> > necessary, merely opening the control and then saying OK with No
> > Secondary Structure Restraints ticked might do it). If you regularize
> > or refine a loop that lacks 2ndary structure with alpha helix
> > restraints turned on you do get a mess.
> >
> > Not sure if that's your problem but I thought it worth mentioning...
> >
> > Cheers,
> >
> > Joel
> >
> >
> >
> > Dear list,
> > Today we upgraded to v0.2 and discovered that regularization seems
> > significantly different from v0.12. I greatly appreciate the ability to
> > add torsion restraints, since I'm trying to use coot on a ~3.5A
> > structure. But regions that regularized stably with v0.12 (matrix =
> > 20.0) get thrown all around the place with v0.2 (matrix = 20.0-50.0,
> > torsion restraints on or off). Thrown around to the point that side
> > chains or carbonyls are ripped right off the backbone -- it's quite
> > ugly. I've noticed that the situation seem to be improved if I'm either
> > trying to regularize a very short poor section surrounded by a lot fo
> > good stuff (2-3 residues seems the limit), or the section is already
> > perfect. But when I try a weird loop that needs significant fixing (for
> > example, 6 residues with 3-5 good ones on either side), regularization
> > eats it alive. I recently saw a message from R. Baxter on the same
> > subject, but no posts about it since the 17th. Are others having this
> > problem? Is there a quick fix that we could implement?
> >
> > Jacob
> --
> http://www.ysbl.york.ac.uk/~emsley
> Structural Biology Laboratory, University of York, York, YO10 5YW, UK
--
Richard Baxter <[log in to unmask]>
