[phenixbb] local refinement of target weights?

Pavel Afonine pafonine at lbl.gov
Tue Oct 4 13:32:58 PDT 2011


Hi Yarrow,

> I am wondering if there is a way to restrain geometry target weights for a
> ligand seperate from the rest of the molecule. I would like to impose
> higher restraints for the ligand than that of the protein in the same
> refinement run.  Thanks for your help.

currently you can only do it by editing the corresponding esds in ligand 
CIF file. By making the esds smaller you will make the corresponding 
restraints stronger.

But in general it is dangerous to do, and normally is not recommended. 
If you make the esds too small, that means the corresponding 
contribution to overall restraints target arising from that ligand will 
be overwhelmingly large, so the restraints from other atoms will be 
negligibly small. In practice this means you will get a model with 
distorted geometry.
This is exactly why often the CIF files created by PRODRG do not work 
well with phenix.refine (since it sometimes puts very small esds, which 
explodes refinement).

Pavel



More information about the phenixbb mailing list