components into density maps to generate complete atomic mod-
els. Software like Situs [18] and MDFF [19] can be used for this
purpose. An extensive review of computational methods for refine-
ment of structural models, including approaches that integrate
experimental data, is given in ref.20.
Essential to the study of drug-bound systems, the initial model
must capture the structural interaction between the target protein
and drug molecule (seeNote 2). Ideally, a crystallographic structure
will be available for their complex. In the case of the example
project for this chapter, the experimental structure already contains
atomic coordinates that describe the drug binding mode. When the
binding mode is unknown, it may be possible to produce a suitable
structural model based on molecular docking. Software like Auto-
Dock Vina [21] can be used toward this end; however, experimen-
tal validation is recommended to support the predicted binding
mode. A review of current molecular docking methodology and
comparison of available docking programs is given in ref.22.
Even when the initial model is based soundly on experimental
data, it is important to perform a series of standard structural
assessments to verify the model’s integrity and to check for subtle
issues that may cause serious problems during the simulation phase
of the project. One critical test is a check for structural clashes or
close contacts within the model. Such deficiencies can exist even in
high-resolution crystallographic structures, and if not corrected,
may cause the model to distort or blow apart during simulation.
Close contacts may also manifest as ring moieties punctured by
neighboring functional groups, or valence angles that are extremely
small. At minimum, these structural issues may introduce instabil-
ities during simulation, requiring increased equilibration time, so
taking care to address them during the model preparation phase of
the project is good practice (seeNote 3). In the case of the example
project for this chapter, no structural clashes or close contacts
within 1 A ̊are found in the crystallographic structure.
Another structural discrepancy to test for is the presence ofcis
peptides in the protein. Typically, the peptide bonds of the protein
backbone adopt a trans orientation (Fig. 3a); however, cisFig. 3(a)Transorientations of peptide bonds are most common. (b)Cisorientations of peptide bonds may be
encountered occasionally. (c) Peptides containing proline can readily adopttransorcisorientations
248 Jodi A. Hadden and Juan R. Perilla