Hi, a few things to keep in mind:
Secondary structure (SS) restraints in refinement programs (at least in Phenix) depend on correct SS annotation that the program either takes from the user input or attempts to do automatically. Quality of automated SS annotation strongly depends on geometrical quality of input model. For poor model you should not expect good SS annotation. When you supply SS to the program then it is entirely your responsibility to do this right. Furthermore, even correct SS annotation does not guarantee that refined structure will obey input SS annotation. This is because if input model geometry is poor enough then refinement may not guide the model over potential barriers so that it adopt requested SS. All this is illustrated here: https://www.phenix-online.org/presentations/ss_idealization.pdf Pavel On Fri, May 29, 2015 at 8:35 AM, Smith Liu <[email protected]> wrote: > Dear All, > > Even with the Coot secondary structure (for example helix) restraint > selected and by this way we keep the secondary structure, I find the > protein secondary structure formed in this way was not so typpical, for > example, not all CO ( i) and NH (i+4) forms H-bonds, and there are H-bonds > formed by CO ( i) and NH (i+3). I checks some RCSB PDBs, and find this kind > of untypical H-bonds for the Helix backbone were not rare. > > Do we accept this kind of untypical H-bonds containing PDB as normal, or > do we think we need further refine the structure based on the map? > > Smith > > > >
