Hi John,

I would like to restrain the positions of ribose/phosphate atoms of altloc 'A' to the positions of those in altloc 'B'. Problem is, altloc A and altloc B are different residue types, so phenix.refine exits stating that there are no common atoms when I enter the following NCS-type restraint:

selection: chain I and altloc A and resid 1 and (phosphate or ribose) restrain: chain I and altloc B and resid 1 and (phosphate or ribose)

I am using the phenix GUI. Is there a better place to enter this type of restraint (rather than NCS) that does not care if a P atom is from a Cd or from a Gd residue type? thanks.

No, sorry. our NCS selections require matching atom labels (resid, name, altloc) except for the chain id. When I implemented the NCS selection I tried to make them more general but it turned out to be quite difficult and in the end I took the "simple and certain" path. I have my doubts about supporting NCS restraints of alternative conformations; the two seem to be mutually exclusive. But maybe I don't understand your situtation. I can see that NCS restraints of backbone atoms only are useful, even if the residue names are different. The trouble is that atoms in two chains need to be matched unambiguously while allowing for chain breaks. I could try to simply remove the requirement that the residue names have to match and go by residue sequence number + insertion code only. Would that work in your case? A potential downside is that people have to be careful excluding non-matching sidechain atoms. Ralf