Mike White wrote: > I believe that one reason that one cannot compare molpdf scores > between models of different proteins is that it depends on the number > of atoms in the protein- more atoms means a higher molpdf, even if it > is a "good" model. Does this mean that if one is trying to do an all > hydrogen model that the molpdfs will be much higher than that for the > same protein modeled without hydrogens?
Yes.
> The reason that I ask is that I am modeling a pentamer with an all > hydrogen model, and each time the modeling stops after a few steps > with the moldpdf stuck slightly higher than 100,000 (which is the > default max_molpdf value) and then moves on to the next model, where > this happens again (and again...). I assume that this is because of > all of the additional hydrogens, and that if I just increase > max_molpdf to 200,000 things should be fine.
You can certainly try increasing max_molpdf; however, this could also be an indication of a bad model, so you should check the results carefully.
The default is to build models without hydrogens, since including them increases computational time roughly 4X and most crystal structures do not include them anyway.
Ben Webb, Modeller Caretaker