[phenixbb] Exploded ligands and phenix...
Ruben Van der Meeren
ruben.vandermeeren at ugent.be
Tue Apr 28 14:36:16 PDT 2009
I assume that it is phenix. Because when I refine with refmac5, the
ligand doesn't explode.
I too had such problems in coot as you describe, but these are fixed
now. The cif file was corrupted. So when I used "real space refine
zone" in coot the ligand break up (what in fact happen, is that coot
doesn't show the bond if it is deviates too far from normal e.g. 2A in
stead of 1.5A, but pymol does which give this kind of strange
structures: http://gallery.me.com/ruben.vandermeeren#100216 -->
"pymol version of exploded ligand").
This problem of the cif file is solved now (using the PRODRG server).
And as Ralf said, this "new" cif file seems to be correct. But is
there a change that phenix cannot read the cif file properly?
Citeren Maia Cherney <chern at ualberta.ca>:
> Hi Ruben
> are you sure that it's a phenix problem? Did you try other programs to
> see your ligand after refinement?
> I have seen a similar problem in coot with DNA. Coot would break up a
> DNA when we tried to regularize it, due to a different letter code
> convention for nucleotides. But when we changed A to Ad, G to Gd etc,
> there were no problems.
> Ruben Van der Meeren wrote:
>> Hi Ralf,
>> The coordinates of the ligand in the file I use for refinement are
>> similar to those I posted. I uploaded an overlay of the original
>> ligand and the ligand in my model. Green is model, blue is original...
>> You can see the picture on this site:
>> The cryst1 card of the pdb is the next:
>> CRYST1 98.500 104.050 415.200 90.00 90.00 90.00 P 21 21 21
>> The structure has 4 monomers in which I modeled a ligand monomer A and
>> B. I also checked the .geo file. In most cases (for bonds and Angles)
>> the model is almost equal to the ideal. For dihedrals there is much
>> more disagreement. E.g. for one dihedral the model is 168° whereas the
>> ideal is -60°. For non-bound interactions the distance is sometimes
>> less than the VDW-radius. Could this be the problem? But this is in
>> fact the thing I want to refine.
>> The log file can be downloaded here:
>> Citeren "Ralf W. Grosse-Kunstleve" <rwgk at cci.lbl.gov>:
>>> Hi Ruben,
>>>> I'm using phenix for the refinement of my protein (phenix.refine). But
>>>> there is a problem...
>>> Your pdb and cif file seem OK after a quick check (phenix.pdbtools
>>> --geometry-regularization tre.pdb tre.cif).
>>> My best guess is that the ligand conincides with a symmetry operation.
>>> Are the coordinates in the pdb file you posted similar to the coordinates
>>> in the file you used for refinement?
>>> Could you send me the CRYST1 card?
>>> The phenix.refine log would also be helpful.
>>> Note that all geometry restraints are listed in the .geo file written
>>> by phenix.refine. Look for nonbonded interactions involving your ligand.
>>> phenixbb mailing list
>>> phenixbb at phenix-online.org
>> phenixbb mailing list
>> phenixbb at phenix-online.org
> phenixbb mailing list
> phenixbb at phenix-online.org
More information about the phenixbb