SHELXL does not provide torsion angle restraints. As far as I can tell, the normal procedure using REFMAC and PHENIX.REFINE is to refine with torsion angle restraints applied to side-chains but not to the psi and phi angles (all three programs have ways of keeping peptide units approximately planar). In my biased personal opinion, the torsion angle restraints on the side-chains are a major reason why these two programs are more suitable for lower resolution refinement (>2.0A) than SHELXL, though there are several other minor reasons. The use of 'antibumping' (VdW) restraints in all three programs in their current rather primitive form does not prevent one from getting plenty of Ramachandran outliers, so the Ramachandran plot is still a good diagnostic tool.
George Prof. George M. Sheldrick FRS Dept. Structural Chemistry, University of Goettingen, Tammannstr. 4, D37077 Goettingen, Germany Tel. +49-551-39-3021 or -3068 Fax. +49-551-39-2582 On Wed, 7 Nov 2007, Garib Murshudov wrote: > I could not say about other program but in refmac there are intra residue vdw > repulsions if atoms are at least two bonds are apart. > If atoms are only two bonds apart then vdw distance is reduced and it can be > controlled. > > I do agree that at lower resolution you should use as much info as you can > get. Problem is how to parameterise info you have. > The main problem with phi/psi is that there is no functional form that can be > easily implemented and would make chemical sense > VDW repulsions take to some degree into account phi/psi but not completely. > You can add restraints explicitly where you want > using for example distance restraints in refmac. See for example > www.ysbl.york.ac.uk/refmac/data/refmac_news.html > > It is only for refmac. I am sure that other programs like phenix.refine, > shelxl and cns have their own implementations to deal with these cases. > > > regards > Garib > > > On 7 Nov 2007, at 18:19, Edward A Berry wrote: > > > Bernhard Rupp wrote: > > > Dear All, > > > I wonder about the exact use of torsion restrains and the effect > > > on phi/psi/w validation. > > ---- > > > a) does refmac restrain phipsi at all (except vdw repulsion > > > which does not bias them otherwise)? > > > > Isn't it the case that vdw interactions are turned off > > between atoms in a residue, and perhaps with the residues > > before and after? > > Otherwise positional refinement would in fact tend to drive > > the residues into the rama-allowed region? > > > > Might it be a good idea to use torsion restraints, or > > intra-residue vdw repulsion, in the last stage of refinement > > after calculating an unbiased ramachandran plot; > > especially in low-resolution problems where we need all the > > restraints we can get? > > > > Ed > > >
