On Wednesday, 10 June, 2015 17:35:25 Mengbin Chen wrote:
> Hello everyone,
> I am refining a 2.5 angstrom structure whose phase is solved by molecular
> replacement with a search probe determined by SAD with 3 angstrom
> resolution. While I am able to see densities of a bunch of water molecules
> and ligands in the MR solved structure, which means that the phase is
> correct, the Rfree gets stuck at ~27%.
Why do you think this is a problem?
It is true that turning on twin refinement is expected to yield lower
R factors, and thus 27% in the presence of twin refinement is worse than
27% without twin refinement. But having said that, Rfree = 0.27 is still
not so horrible.
If the starting point for your MR solution was a 3A model,
there may have been errors in sidechain placement or even
backbone conformation that can be corrected now that you
have higher resolution data. Don't assume that the starting
model was perfect.
You say that you can see density for a bunch of waters and ligands.
Are these in your refinemed model yet?
What does Molprobity (or coot or phenix) say about the quality of
your sidechain rotamers?
Have you tried adding a TLS description?
> The crystal belongs to P3221 and has
> a twinning fraction of 19%, according to Xtriage. Currently I've been
> sticking to Phenix for refinement, and twin law (-h, -k, l) has been
> I was wondering if the CCP4 community would have any suggestions of how to
> refine this twinned structure, such as softwares to use, tricky strategies
> to choose, etc. I really appreciate any recommendations you would come up
> with my situation!
> Thank you in advance!
Ethan A Merritt
Biomolecular Structure Center, K-428 Health Sciences Bldg
MS 357742, University of Washington, Seattle 98195-7742