Just a general observation to stir up the discussion, as I'm eager to hear
the opinion of others on this important topic.
I think a 4.4A map is still rather low resolution to allow for a
well-defined refinement of the atomic model (e.g. where angles remain in
the allowed regions of the Ramachandran plot without proactively forcing
them in there), and this likely holds for most refinement programs out
there. The 3.5A in the best region of your map may be good enough, but at
resolutions lower than 3.8-4A things start becoming much more difficult as
side chain densities disappear and it becomes less clear where the
, which may elicit some discussion, and I would be .
> Hi all,
> I am building / refining a model on a ~4.4A map using coot, and then use
> phenix.real_space_refine for refinement, which cleans up the clashes,
> and bond length/ angles nicely, but somehow screws up the Ramachandran
> plot. I guess part of the reason could be that the resolution isn't
> uniform across the whole EM map, (3.5 to 5A in my case), so that may
> confuse the program about the b-factor distribution in the map.
> I guess Refmac is the refinement program that many people here are using,
> but I am curious on what other refinement program is popular, and how
> does the real space refinement compare with reciprocal space in refining
> models in EM map. And what tool to use for converting a mrc map into
> reflection file for reciprocal space refinement like Refmac?
> Thanks a lot.
MRC Laboratory of Molecular Biology
Francis Crick Avenue, Cambridge Biomedical Campus
Cambridge CB2 0QH, U.K.
tel: +44 (0)1223 267061