[phenixbb] Low resolution structure

Nathaniel Echols NEchols at lbl.gov
Thu Mar 11 13:02:43 PST 2010

On Mar 11, 2010, at 3:40 PM, Kendall Nettles wrote:
> My second question is whether someone can suggest appropriate distance_ideal,    sigma, and  slack for hydrogen bond restraints for a helix? Also, I have a 2 angstrom structure of this part of the complex for reference. 

I was hoping someone would ask this question.

If you download version 1.6.1-336 from the nightly builds page, phenix.refine has a very preliminary implementation of secondary structure restraints that I'd like to test.  To use, add "main.secondary_structure_restraints=True" to the command line.  In theory, everything else is automatic, but you may need to tweak some of the parameters in the secondary_structure scope to get it to work properly.  If you want more control over secondary structure assignments, run phenix.secondary_structure_restraints to generate the group selections for use in phenix.refine.  (You can do all of this in the GUI too, but it is deliberately kept hidden until I'm sure it works - I can explain how to find it though.)

The parameters that I found worked best were:

1.975 A for distance_ideal
0.02 for sigma
0 for slack

although the default for sigma is currently 0.05.  I found it worked much better to use actual hydrogens for this, but you can fake it with N-O restraints with a distance of 3A.  The default behavior is to throw out any "bonds" that exceed 2.5A, because the annotations are often unreliable and can improperly restrain atoms, but if your starting structure is particularly messy you may want to turn filtering off ("h_bond_restraints.remove_outliers=False").

Let me know if this works; I haven't had much opportunity to test it thoroughly, but it can improve R-free even at moderate resolutions.  (It can also hurt, but this appears to be less common so far.)


Nathaniel Echols
Lawrence Berkeley Lab
NEchols at lbl.gov

More information about the phenixbb mailing list