There may be several reasons:
It looks strange. Something seems to be wrong in protocols
1) Check free R flags. I have seen in few cases wrong free R.
Generally number of reflections for free should be less than for work
reflections
2) Check TLS refinement. You may need to restart TLS refinement. At
this resolution I would try anisotropic refinement. If your output pdb
file has aniso card then it may be the resason. Refmac by default uses
isotropic refineemnt and removes all aniso cards. You can turn on
aniso refinement
3) Check hydrogens. Adding hydrogens in their riding position may make
difference
4) Check twin refinement
5) Anisotropic refinement
If above reasons are not applicable to your case then you can send the
data and I can check what is wrong.
regards
Garib
On 13 Jul 2010, at 20:32, Ariel Talavera wrote:
Hi CCP4ers,
Is there any differences in the way of calculating the Rfactor between
PHENIX.refine and other programs implemented in CCP4i, like Refmac5 or
sfcheck, or the EDS server?
I asking this because I have a structure solved at 1.4 Å resolution.
Phenix,refine was used for its refinement and for the final model
Rfactor was 14.5%. But during the validation with sfcheck (CCP4i), I
got
an Rfactor of 20.8%. The same happened when I used this final model as
starting model for a refinement with Refmac5. Again I got an Rfactor
of
20.8%.
As far as I know, Phenix uses different algorithms for the refinement
than other programs, which in some cases can make Phenix gets better
Rfactors. Could these differences be the reason for this large
difference in the calculated Rfactors? Or I have to recheck my
procedure
for mistakes?
Thanks a lot in advanced.
Looking forward to hearing form you.
Cheers,
Ariel