Bo Yang wrote: > The models built from Modeller 8v1 contain some residues (non-Gly and > non-Pro) in the disallowed region based on Ramachandran plot. I wonder > how this could happen? How to change these residues back to allowed region?

Sure, this can happen when Modeller builds the model, since it tries to satisfy all the restraints, and while there are dihedral restraints that try to keep residues out of the disallowed regions, these can conflict with other parts of the system - for example, a steric clash with some other part of the backbone or sidechains. You can try to refine these residues in another package if you like, or add explicit restraints on them to try to force their conformation into what you feel is the "right" one, but you can't guarantee that will work, of course.

Ben Webb, Modeller Caretaker