[phenixbb] Question about antibumping restraints and restraints on dihedrals

Ralf W. Grosse-Kunstleve rwgk at cci.lbl.gov
Wed Aug 22 04:54:52 PDT 2007


Hi Madhu,

> 1. Is there a way of applying antibumping restraints to decrease the
> number of clashes in phenix.refine, I looked at older posts but couldn't
> get a satisfactory guideline regarding the same.

The antibumping restraints (nonbonded interactions) are on by default.
Currently there isn't a syntax to turn them off or to change the
default parameterization.
If you refine for a few cycles with the weight for the x-ray target
set to zero (wxc_scale=0) the clashes should definitely go away.
Then go back to normal refinement (wxc_scale=0.5). I'd hope that
the clashes don't come back. Otherwise that's what the x-ray data
guide the minimization to, even though the restraints are pushing
back.

> 2. The default setttings for geometry refinement looks like this:
> refinement.geometry_restraints.remove {
>   angles = None
>   dihedrals = None
>   chiralities = None
>   planarities = None
> }
> Do I have to change "dihedrals=True"

This will not work:

Sorry: Invalid atom selection:
  refinement.geometry_restraints.remove.dihedrals="True"
  (RuntimeError: Syntax error.)

You have to specify atom selections to *remove* default restraints.

> if I do not want to put restraints on
> dihedrals. I was thinking that one puts restraints on psi/phi angles only
> by setting "discard_psi_phi = False" because when I set it to
> "discard_psi_phi = True", the Ramachandran plot was much better. But I did
> not want to do this because then the Ramachandran plot is no more a
> validation tool. I just wanted to know if Unknowingly I did that by
> setting up "dihedrals = None".

No, if you leave the geometry_restraints.remove block untouched
you'll get all restraints incl. dihedrals, except for the peptide
psi and phi angles, which are controlled by the dedicated parameter
you quote.

Cheers,
        Ralf



More information about the phenixbb mailing list