Small Molecule Docking from Theoretical Structural Models - Computational Modeling of Biological Systems - page 80

Fig. 2 Number of citations in scientific literature of commonly used docking algorithms in 2009

and are never published, analysis of the literature reveals that the word “docking”
has been used in the title or abstract in 1,565 publications during 2009.
Docking can be done in quite different scenarios, where objectives and success
criteria can be quite different:

1. Derivation of structural binding mode for a known binder
2. Determination of primary or secondary targets for a drug
3. Virtual high-throughput screening (vHTS)
The derivation of a structural binding mode for a small molecule is probably the
most traditional use of docking algorithms. Within this paradigm, the process starts
after high-throughput experimental studies (or alternative methods) that detect one
or several small molecules which display activity against a given target. However,
there are many factors that determine whether these “hits” can become “leads”
or can be modified to improve their properties. Such a lead optimization process
requires a quite detailed knowledge of the binding mode, something that only in
silico docking can provide with the required velocity. In this context, the use of
docking methods is defined by the limited number of drugs to consider and by the
existence of a single target protein. The accuracy is, however, crucial since errors in
the placement of the drug can completely misguide the lead optimization process.
A basic metric commonly used for evaluating the accuracy of the predicted binding
modes of docking programs is the root mean square deviation (RMSD) between the
predicted conformation and the native pose of the ligand:

X ! 1=2

R j / 2
N

.R i

RMSD D ; (2)

N

where R stands for the ligand coordinates in the predicted binding mode .i / and
in the native pose .j /,andN is the total number of atoms. In many practical