saccenti@cerm.unifi.it wrote: > Which is the best strategy to assess the quality of a model and sort them > out of a bunch of many calculated models? > > I have models a complex of two proteins plus a metal ion.

Of the methods we have available in Modeller, DOPE is probably the most reliable - although as a statistical potential, it won't score your metal ion at all. See http://salilab.org/modeller/9v2/manual/node29.html for an example.

Ben Webb, Modeller Caretaker