On Monday, October 25, 2010 03:16:28 pm Maia Cherney wrote:
> I had a similar problem. It dissappeared when I switched the
> refinement to phenix. The R factors dropped and the difference between
> them became acceptable.
It would be useful to pin down what the two programs did differently.
It is likely that some particular default setting is the relevant quantity.
- Geometric restraint weight?
- B-factor restraint weight?
- NCS?
- hydrogen treatment?
- Did you use the same set of R-free reflections in both programs?
I'm currently running a bunch of parallel runs in phenix and refmac
to test something else. What I notice to be the biggest difference
in their default settings is that the default restraint on B factors
is looser by a factor of 2-3 in phenix, whereas the default restraint
on geometry is looser by a factor of ~2 in refmac. I need to adjust
those for each refinement pair so that I can get comparable runs.
Ethan
--
Ethan A Merritt
Biomolecular Structure Center, K-428 Health Sciences Bldg
University of Washington, Seattle 98195-7742
|