*** For details on how to be removed from this list visit the *** *** CCP4 home page http://www.ccp4.ac.uk ***
Bill, I'm hesitant to make suggestions to someone with as recognizable a name as yours for fear of sounding stupid, but two things have come to mind. First, is there something similar to the CNS anomalous scatterer libraries ( input of f' and f" ) that is necessary for refinement in Refmac to properly model the Mn? Second, I have recently been convinced by some collegues in the Richardson lab at Duke that the pucker restraints in CNS should be used. The reasoning for this as I understand is that in high resolution structures of RNA/DNA the sugars almost exclusively fall into a well defined set of puckers. With most RNA sugar puckers being either 2'-endo or 3'-endo. Lowering the dihedral restraints in CNS or using Refmac will pull the sugars back into density and reduce the Rs, but it may not realistically model the RNA backbone. In my experience, which is limited to the re-refinement of only one structure, you get a greater decrease in Rfree if you first define the puckers as either 2' or 3' before relaxing the dihedral restraints. The Richardson's have tools on the molprobity website that help to identify possible sugar pucker problems and unlikely base-phosphate perpendiculars. Also, using CNS with defined puckers forces you to confront and individually inspect problem sugars, but Refmac in my opinion smooths out, without always fixing these same problems. But to be fair, this whole argument favoring defined puckers falls apart at high resolution because the density alone becomes sufficient. Sorry for the longwindedness. Best Wishes, -bob
