On 10/21/24 2:28 AM, Marco Sette via modeller_usage wrote: > I have obtained five models of a protein with Modeller and I am > analyzing them in relation to the wt. There are of course small > deviations, and such differences affect some physical properties. In > your opinion, should I analyze the separate models or I can take the > average structure? In NMR, people often use an average structure so I > wonder if it makes sense in the modelling field.
Normally we look at the individual models. If you do want to average the structures, I would recommend you at least relax the final structure. One way to do that is to use AutoModel.cluster() as per https://salilab.org/modeller/10.5/manual/node71.html which does a basic conjugate gradients minimization.
Ben Webb, Modeller Caretaker