Choice and preparation of the starting structure

As with any protein simulation, the nature and limitations of the structural solutions for proteins provided by X-ray crystallography should always be borne in mind [125]. One obvious point is that hydrogen atoms are generally not observed because of their low electron density (neutron diffraction experiments can be useful to overcome this problem), and so it can be difficult to assign protonation states unambiguously, and to decide between possible rotamers or tautomers. This, and other factors such as model bias (for example in a molecular replacement solution), or simple error in construction of a model, may lead to the structural model being incomplete or incorrect in some places. 

It should be remembered that a crystal structure represents an average not only over all the molecules in the crystal, but also over the time course of data collection. Dynamic and static disorder can lead to weak electron density in some regions, complicating the building and refinement of a structural model [9,125]. Caution should be exercised, and such factors considered, when a 

It is very important to assign protonation states of protein groups correctly. 

---