On Tue, Jun 22, 2010 at 2:21 PM, Scott Classen <sclassen@lbl.gov> wrote:
Hello friendly Phenix developers,

According to the CHANGES for phenix.refine 1.6.2-432 there is a new feature:

"hydrogen-bond restraints for Watson-Crick base pairs"

How do I use this? I couldn't find any documentation... or is it automagic?

Automagic, I hope.  Start with "main.secondary_structure_restraints=True", and it will attempt to find existing base pairs by analyzing hydrogen bonds.  If your geometry is still a little screwy, the parameter syntax is like this:

refinement.secondary_structure.nucleic_acids {
  base_pair {
    base1 = chain "A" and resseq 1
    base2 = chain "B" and resseq 10
  }
}

One piece of advice: make sure your structure has either all hydrogen atoms where they should be, or none at all - if you're missing any, the secondary structure restraints will probably break.  The automatic mechanism will also restrain protein helices and sheets, but you can easily turn this off (restrain_helices=False restrain_sheets=False).  It will probably be necessary to fiddle with the sigma and slack parameters for hydrogen bonds; it isn't clear yet how much these need to be tuned for individual structures, but the defaults don't always work best for all structures.

I am working on a brief sort-of-article on the secondary structure restraints, which has more information about the general mechanism (currently more focused on proteins, which is perhaps a defect).  I'll send you a copy of the current draft.  (A modified version of this will eventually become the official documentation.)

-Nat