Dear Brian,
you could also manually edit the cif files to reduce certain restraints. I sometimes find it useful to weed out excessive numbers of dihedral and torsion restraints that have been generated automatically. Also if you explicitly read in a cif file, it usually overrides existing parameters. With heavy atoms like Fe and S, you should be able to relax the restraints quite a bit and let the diffraction data decide where the atomic positions are.
Best,
Herman
Von: CCP4 bulletin board [mailto:[log in to unmask]] Im Auftrag von Brian Eilers
Gesendet: Freitag, 24. Juli 2015 18:11
An: [log in to unmask]
Betreff: Re: [ccp4bb] Metal cluster refinement.
Jens,
That is what I was afraid of, I have tried to make my own with limited success, refmac was having trouble reading the cif that I created, I would appreciate it very much if you could send me the cif files that you have used in the past. Yes it would be helpful it you could just use one file for both forms since these iron sulfur clusters are not constrained as much as the dictionary files suggest.
Ethan,
I have tried using both dictionary files since I am trying to see if the P-cluster is oxidized or reduced, and both work seem to give the reduced form or even tighten up the reduced form more than it should.
Would another program besides refmac be more appropriate? CNS?
Thanks,
Brian
On Thu, Jul 23, 2015 at 11:41 PM, Ethan Merritt <[log in to unmask]<mailto:[log in to unmask]>> wrote:
On Thursday, 23 July 2015 09:10:50 PM Jens Kaiser wrote:
> Brian (and PDB staff listening in),
> Restraint files downloaded from the pdb are autogenerated, you
> unfortunately can't even deposit cif files alongside your coordinates
> and diffraction data.
That does not match my experience.
cif files for ligands, at least for ligands with a novel name,
are not only allowed they are requested [required?] as part of the
deposition.
I have the impression the deposited cif dictionary is used for validation,
but it would be nice to have confirmation of that.
It would also be nice if it were also carried through and attached to
the archived PDB entry so that the refinement can be replicated.
> And autogeneration does not work well for metal
> clusters. Based on connectivity/atom names, the "ox" and "red" forms of
> the P-cluster are identified as the same compound. The correct treatment
> should be to have a constraint file with both geometries and the same
> ligand-ID, but that's not possible to implement. Your best bet is to
> create the restraints by hand using one of the high-res structures as a
> guide, and give the ox and red different names during your refinement.
> The name will be changed during deposition.
The current PDB ligand dictionary has code
1CL for the oxidized form (present in 2 entries)
and code CLF for the reduced form (present in 24 entries).
Both are present in entry 3U7Q.
The two dictionary files are definitely not the same, although I
did not pursue it any further than putting them side-by-side.
I do not know whether either of these would be a correct starting point
for the new structure being refined.
Both are listed as having a formal charge of 0, which doesn't
sound promising.
cheers,
Ethan
> It's a pain, I've gone
> through it a couple of times. If you want, I can send you cif files we
> used in the past off-list.
>
> HTH,
>
> Jens
>
> On Thu, 2015-07-23 at 16:21 -0600, Brian Eilers wrote:
> > Hello,
> >
> >
> > I am trying to refine a protein structure that has a Fe(8)-S(7)
> > cluster bound to it. I have been using a cif file that has the ligand
> > 1CL downloaded from the Protein Data Bank. This cluster is in the
> > oxidized form and should be that way in my structure as well, but when
> > I use refmac in CCP4i (6.4.0) it ends up refining the cluster to the
> > reduced form. The maps have positive density where the cluster should
> > be, indicating that the cluster should be in the oxidized form. It
> > seems like refmac wants to idealize the bonds in the cluster instead
> > of using the map as a guide for where the Fe and S atoms should go. Is
> > there a way in refmac to ease the restraints on the cluster so that it
> > refines with the data, and not make idealized bonds?
> >
> >
> > Thanks,
> >
> >
> > Brian
> >
> >
> > Brian Eilers
> > Research Associate
> > Dept Chemistry and Biochemistry
> > Montana State University
--
mail: Biomolecular Structure Center, K-428 Health Sciences Bldg
MS 357742, University of Washington, Seattle 98195-7742
