On Mar 16, 2010, at 9:32 AM, Paul Emsley wrote:
Coot needs to be told about the geometry of the link so that it can refine it.
Why wouldn't it be sufficient for Coot to approximate/guess these parameters for the purposes of building? No one is going to do an all- atom refinement of their structure in Coot (are they?). That is the domain of downstream programs like refmac. It seems reasonable to allow an approximated bond to be made in Coot so that the user can build and then let the full refinement fail in the downstream program when the proper topology descriptions are missing for the manual bonds. Or is the fear that the user will forget he added new bonds and that said ligand will diffuse away inexplicably during refinement?
James
