Chemistry Reference
In-Depth Information
for a protein target many putative binding conformations/orientations
at the protein's active site. A scoring function is then used to rank the
ligand orientations. The experimental binding mode would be of
course more highly ranked by an ideal scoring function.
b)
Prediction of absolute binding affinity between ligand and protein. In
lead optimization, the strength of binding affinities for selected
compounds (lead compounds or low-affinity hits) is improved.
c)
For a given protein target, identification of potential drug/leads
via
screening of databases. Known binders are ranked highly by reliable
scoring functions. Virtual database screening, binding affinity
prediction and ligand binding mode identification are interconnected
such that equal performance for all three applications are obtained by
an accurate scoring function.
We can also group the scoring functions in a) force field-based, b) empirical,
c) knowledge-based.
a)
The first type,
i.e.
FF (force field scoring functions) are based on
physical chemical like atomic interactions. These include electrostatic,
van der Waals (VDW) interactions as well as torsional/bond
stretching/bending forces. The functions and their parameters can be
obtained from
ab initio
quantum mechanical calculations or/and
experimental data. The solvent is a challenging problem. The
Lennard-Jones VDW and the electrostatic terms of the force field
energy components (1/єr
ij
2
as well as (r
ij
-6
), (r
ij
-12
) pairwise distance
dependent terms respectively.
The effect of the solvent is implicitly introduced by a dielectric
constant, which is dependent on distance. This dielectric factor cannot
account for desolvation effects because non-polar groups tend to
remain in non-aqueous environment and charged groups favor
aqueous environments. Desolvation energy depends on specific
geometric and chemical environments of solute atoms. It is a many
