Okay, this is what I would expect. The torsion NCS restraints are
designed to not restrict atoms that are in very different positions,
so in a case of incorrect nomenclature the restraint would simply not
be active.
We are working on some ideas for correcting nomenclature (or at least
flagging such instances) which should hopefully make it easier to deal
with such problems when they arise.
Jeff
On Fri, Jan 20, 2012 at 10:53 AM, Luca Pellegrini
Hi Pavel,
NCS was set to 'global' in the refinement gui. Setting it to 'torsion' does seem to maintain the correct geometry.
Luca
On 20 Jan 2012, at 17:20, Pavel Afonine wrote:
Hi Luca,
what kind of NCS restraints did you use: torsion or Cartesian? I wonder if torsion NCS would naturally avoid this problem? I hope Jeff comments on this.
Pavel
On 1/14/12 10:51 AM, Luca Pellegrini wrote:
I think I discovered what was happening to my DNA molecule during refinement, that caused distortion of the phosphate groups. It was something along the lines of what Dale suggested, I mention it here so that it might be useful to others.
I have two DNA molecules in the ASU and I didn't realised that I had accidentally swapped OP1 and OP2 in 3 out of 28 phosphates (so that, if the two DNA molecules were superimposed, OP1 ended on top of OP2). Because I had NCS restraints on during refinement, I guess phenix was trying to move OP1 in one DNA molecule towards the NCS-equivalent position of OP1 in the second molecule, hence the altered geometry.
I have fixed the atom nomenclature problem and now the geometry of my DNA is ok. I appreciate that most people tend to avoid such mistakes ;-) but perhaps it would be good if phenix could flag situations such as these.
Luca
On 6 Jan 2012, at 16:15, Pavel Afonine wrote:
Hi Luca,
phenix.refine always generates a *.geo file that lists all the geometry restraints (bonds, angles, planarities, chiralities, dihedral, non-bonded, ncs(if any)) used in refinement for all atoms. For each restraint it list current model value, target (library) value, etc.
Can you have a look at *.geo file for O-P bonds in question? May be this gives a hint about what's going on?
Pavel
On 1/6/12 4:31 AM, Luca Pellegrini wrote:
Hi,
I have noticed that Phenix mangles O-P bond lengths and angles in the phosphate groups of my DNA chain (example in pict attached). This only happens to 3 out of 28 phosphates. The refined protein-DNA structure does not present any other unusual stereochemistry issues and refinement runs normally. Could anybody advice on what is going on and how to fix this, please?
I am using Phenix-1.7.3-928. Refinement is with default target weights, hydrogens on and does not include simulated annealing. I could make the stereochemistry targets more stringent, but the default values seem strict enough (bond and angle rmsd is 0.005 and 1.225).
Thanks, Luca
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
Luca Pellegrini Department of Biochemistry University of Cambridge 80 Tennis Court Road Cambridge CB2 1GA - UK
Email: [email protected] Tel: 0044-1223-760469 Fax: 0044-1223-766002 Sanger building, room 3.59
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb