Dear all, I was refining complex structures with a DNA oligo that have some noncanonical bent regions in the middle, the resolutions are fairly good, around 1.8-2.2A. I used both CNS and Refmac. I found (not surprisingly) that with Refmac, I got a lot of unusual sugar puckers like C1'-exo that is neighboring the "usual" C2'-endo for B-DNA even in the normal B-DNA region. from reading past emails in the CCP4 archive, it seems Refmac5 doesn't restrain sugar pucker while CNS punishes violations to normal sugar pucking more than violations in bond lengths. I would like to get people's current opinion about which is the better program to use to refine nucleic acid structures? Refmac or CNS combined with modified parameter file that includes alternate common sugar puckers(e.g. C3'-endo)?
Also, I have a naive/beginner's question on Refmac---since I have a modified base in the structure, I had to provide Refmac with a dictionary file of that base, but the phosphate joining this modified base and the next has strange geometry. I know how to restrain the link bond length, but how does one restrain the angle for the linking bond? Thank you very much for your input/suggestions. Best, Melody
