[phenixbb] phenix refinement question

Nathaniel Echols nechols at lbl.gov
Mon Nov 15 13:33:33 PST 2010

On Mon, Nov 15, 2010 at 1:09 PM, Katarina Moravcevic
<katarinm at mail.med.upenn.edu> wrote:
> Hi all,
> I have only started using PHENIX and have made an interesting observation.
> If I run my model through PHENIX refine (individual sites, individual ADPs,
> occupancies, optimized X-ray/stereochemistry weight, optimized X-ray/ADP
> weight options selected) I obtain R/Rfree 0.15/0.18 and very nice geometry
> values (data is to 1.7A). However, if I now take this refined model and run
> it through Refmac (without any refinement) I obtain R/Rfree of 0.19/0.21.
> Can anyone explain why is there such a big difference? Will this be an issue
> during PDB submission?

Take a look at what phenix.refine reports for the number of
reflections used for refinement in the logfile - I believe it will
discard reflections that are flagged as suspicious (can't remember the
reference, I think it's one of Randy Read's papers), and unless you're
using a very recent version, it may also be ignoring reflections where
F=0.  There are probably a half-dozen other reasons why the programs
disagree, but these are less obvious and potentially much harder to

If you're concerned about reproducibility, run phenix.model_vs_data
with the reflections and PDB file, and make sure that the statistics
it reports agree with what you're sending to the PDB.  The program is
well-documented and mostly open-source, and is a good sanity check for
data in the PDB.

The PDB is using a much more primitive method to check R-factors, and
after they process your structure you'll get an email that includes
something like this:

Structure factor validation

                    High_Res Low_Res Compl  Num_Ref  R_obs   R_work
R_free  Corr(Fo-Fc)
Reported PHENIX      2.500    20.005  81.28  15118    0.2183  0.2150
0.2884  N/A
SFCHECK without TLS  2.50     19.64   87.9   8004     0.2850  0.2850
N/A     0.8546
REFMAC without TLS   2.500    67.386  87.609 8937     0.234   0.2336
0.0000  0.918

The lack of TLS in their validation means that they'll even
overestimate the R-factors of most structures refined by REFMAC!
(Although in this case I think I left the ANISOU records in the
deposited PDB file, so I'm not sure what they're doing wrong.)  They
will never complain about this, in my experience, so it's not
something you need to worry about.


More information about the phenixbb mailing list