[phenixbb] phenix distorts ligand

[Mustafa Koksal]

Hi,
I have a problem in phenix refinement at 2.25 A resolution, data is 80% 
complete. I have protein, water, and ions in the .PDB file, and a ligand 
molecule that is distorted during refinement.

I generated the pdb (with H atoms) with ChemDraw3D, edited the atom 
names etc. I used phenix.elbow to generate the .CIF file.

Here is the first problem:
Ligand has trans double bonds, which are written out by phenix.elbow as 
single rotatable bonds although I had correct connect records and H 
atoms in the .PDB file.
I ran phenix.refine, it distorted the ligand, making double bonds not 
planar and nearly cis.

Here is what I have tried:
I edited the .CIF file to change these torsion agles to be CONSTANT and 
180 or 0 deg.

Here is the second problem:
phenix.refine still distorted the ligand.

Here is what I have tried:
I used optimize_wxc/wxu in phenix.refine to optimize the weight, to see 
if I overweighed the data.

Here is the third problem:
phenix.refine still distorted the ligand.

Is there a way to solve this problem? Is there a way to tighten the 
constraints for the ligand only? I still do not understand why it even 
cannot keep the double bonds planar (and trans). I do not want to move 
back to good old CNS...

Thanks to all,
Mustafa

-- 
Mustafa Koksal
Postdoctoral Fellow
University of Pennsylvania
Department of Chemistry
Roy and Diana Vagelos Laboratories
Office: 2090 IAST
Office Phone: (215) 898-2227
Mobile Phone: (267) 319-6443
Mail Address: 231 S. 34th St, Philadelphia 19104, USA
--
"If you grant me a pen, I shall destroy swords and spears"
Mawlana Jalal-ad-Din Rumi
--