 |
 |
I did that in several cases. At 3.5 -4.5 A you can end up having a real mess. To check the usefulness, I run refinement with a low x-ray weight and a high anneal temperature; if the R-free drops, it is a clear sign that it helps you. I found that it depends in the particular dataset and stage if it is useful. the attached input file has additionally an option for B-factor sharpening and the B-factor targets are increased. Play with rama-energy. And one note more, if your model is far off, this makes things worse, because you fall in the wrong minima of the ramachandran and it's harder to get out with the rama-potential enabled!! so your model has to be already quite correct for this to work well. Cheers, Jens On Thu, 2008-04-17 at 15:37 +0200, Mario Milani wrote: > After many warning messages, I think I have to clarify the meaning of > my question. After careful refining, in a low resolution structure > often many residues (mainly in loop regions) can still adopt > different dihedral conformations which are equally compatible with a > low resolution electron density map. Thus, during the final steps of > the refinement, it could be useful to restrain the dihedrals of such > residues using the theoretical informations contained in Ramachandran > plot to support/complete the poor information displayed by the > experimental electron density map. > If this approach is used during the final stages of refinement i think > it could provide structures of better quality. In this context i think > that the use of Ramachandran restrain would follow the basic > principle: enhance the weight of prior knowledge (geometrical > costrains/restains) when experimental data are scarce. > > > mario > > Mario Milani, PhD > University of Milan > Dep. of Biomol. Science and Biotech. > Via Celoria 26, 20133 Milan, Italy > T. +39 0250314892-4 > F. +39 0250314895 > http://digilander.libero.it/mario.milani/ > http://users.unimi.it/biolstru/Home.html > > > > > > > >