Unconstrained optimization of protein geometries

The result of an unconstrained geometry optimization is a stationary point on the potential energy surface. This point has several very important properties. Among these are:

The geometry refers to a specific system, e.g., the starting point for a reaction step or a salt bridge or a neutral system.
Its heat of formation can be related to heats of formation of similar systems (by "similar" is meant systems with exactly the same formula and charge, and calculated using the same conditions, i.e., PM7 and EPS=78.4).

Because of the importance of stationary points, the corresponding ARC files should be properly labeled and stored for future use.

Also, because these structures are so important, unconstrained optimization of protein systems should be carried out with unusual care. Use solvation for all systems of this type, i.e. use EPS=78.4. Optimize the geometry. This might take CPU days or even weeks of effort, so to be on the safe side, use T=4W DUMP=2D. If the optimization is not exhaustive subsequent geometric operations might compromise any results - changes in heat of formation would be a mixture of the original structure still changing and thus its energy decreasing, and the energy change caused by the current geometric operation.

A useful strategy is to optimize the geometry, then if the gradient norm in the ARC file is large, over about 0.2 kcal mol^-1 Å^-1 per atom, simply re-optimize the geometry. To do this, either run the ARC file without any editing, or edit the ARC file to make a new MOPAC data-set, then run that.