On Tue, Mar 5, 2013 at 11:55 PM, Pavel Afonine
Also note that geoemtry terms in phenix.refine are not designed for simulations, they are very basic just enough to make refinement work well, and almost certainly not suitable to do simulations purely based on them. For example there is no electrostatic terms, which means if you don't carefully specify secondary structure restraints your model will unfold.. etc.
Two further points: 1) If you're trying to preserve a crystallographic trimer anyway, one option is to set wxc_scale to a small but non-zero number, i.e. essentially defining an envelope around your protein with the fake data. This will allow your model geometry to refine but limit how far it can move. 2) The reference model restraints could also be useful here (with or without "data"). -Nat