[phenixbb] metal ions in phenix.ensemble_refinement

[Tom Burnley]

Hi Scott,

Couple of things you can try:

1) Harmonic restraints
This will restrain the selected atoms to their starting positions.
Selection works with usual phenix selections e.g.

harmonic_restraints.selections = 'resname ITD or resname OLC or resname OLA'

If you want to tighten them up reduce the 'slack'.  This is the
distance the atom can move in Angstroms from the starting coordinates
before it gets restrained (they are flat bottomed restraints) e.g.:

harmonic_restraints.slack = 0.5

Or increase the weight e.g:

harmonic_restraints.weight = 0.01

2) Add metal ions restraints
Use phenix.ready_set to generate metal ion restraints:

phenix.ready_set my.pdb

This should generate .edits file with the additional metal ion
restraints.  Check they make sense and give that to
ensemble_refinement when you start your runs:

phenix.ensemble_refinement my.pdb my.cif my.mtz my.edits

3) Check the occupancy
Are you sure that the occupancy of the zinc ions in correct?  If they
are given as 1.0 in the input PDB but are actually a little lower then
the simulation will over sample to over come this.  From the standard
refinement check the zinc B-factors are similar to the surrounding
protein atom B-factors.  Small errors in occupancy have a pronounced
effect.  You could try refining the occupancies in phenix.refine.  It
may be unstable at this resolution, if so try running a few parallel
ensemble runs with different occupancies, e.g. 0.95, 0.9, 0.8,...  You
need to edit the input PDB to do this, either manually or, better
still, using phenix.pdb_tools

Hope that helps,

Tom

On 7 February 2014 14:04, Scott Horowitz <horowsah at umich.edu> wrote:
> Hi, I'm trying to use phenix.ensemble_refinement, and my structure has
> several zinc atoms, both coordinated to amino acids and free imidazoles, and
> sometimes both. It looks as though it's not really understanding that the
> zincs really should stay more or less in their coordination spots, and are
> spending some time in kind of weird spots. I couldn't find anything
> documented that seems like it will help, unless there's some clever way to
> use the harmonic restraints for this that I haven't figured out.
>
> And while I'm at it, I have another potentially related problem: the rfree
> of my final ensemble (0.27) is substantially higher than my input structure
> (0.24). And yes, I had been doing my refinement in phenix, with hydrogens
> added, and I optimized pTLS. If you're curious, this is a 1.9 A structure.
>
> Any help and/or suggestions would be greatly appreciated!
>
> Thanks,
> Scott
>
> --
> Scott Horowitz, Ph.D.
> Research Associate
> Howard Hughes Medical Institute
>
> University of Michigan
> Department of Molecular, Cellular, and Developmental Biology
> Bardwell lab
> 830 N. University Ave, Room 4007
> Ann Arbor, MI 48109
> phone: 734-647-6683
> fax: 734-615-4226
>
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb
>