On 1/18/13 7:40 AM, sdimicco wrote: > I have built a membrane protein by Modeller. I am aware that none of the > Modeller assesment methods are likely to work well on membrane > proteins.However, could I use one of the assesment methods to filter > the best models among my results, and veluate the selected best models > with other methods?

If an assessment method does not work well, this means that it cannot reliably discriminate good models from bad, so you can't use it as you suggest.

Ben Webb, Modeller Caretaker