Reviews in Computational Chemistry Volume 18
Reviews in Computational Chemistry Volume 18
Reviews in Computational Chemistry Volume 18
- No tags were found...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Application of Scor<strong>in</strong>g Functions <strong>in</strong> Virtual Screen<strong>in</strong>g 63<br />
Rank Order<strong>in</strong>g Sets of Related Ligands<br />
For structure prediction, structures of prote<strong>in</strong>–ligand complexes from the<br />
PDB can serve as a common pool to test scor<strong>in</strong>g functions. It is more difficult<br />
to draw valid conclusions about the relative performance of scor<strong>in</strong>g functions<br />
to rank order sets of ligands with respect to their b<strong>in</strong>d<strong>in</strong>g to the same target.<br />
First, there are few published studies <strong>in</strong> which different functions have been<br />
applied to the same data sets. Second, experimental data are often not measured<br />
under the same conditions but collected from various literature references.<br />
The latter practice can have especially dramatic effects when<br />
<strong>in</strong>hibitory concentrations for 50% reduction of a biological effect (IC 50 data)<br />
are used <strong>in</strong>stead of Ki values.<br />
On average, empirical scor<strong>in</strong>g functions seem to lead to better correlations<br />
between experimental and calculated b<strong>in</strong>d<strong>in</strong>g energies than do force field<br />
based approaches because the nonbonded <strong>in</strong>teractions <strong>in</strong> the latter are usually<br />
not optimized to reproduce <strong>in</strong>dividual <strong>in</strong>termolecular b<strong>in</strong>d<strong>in</strong>g phenomena.<br />
However, the only available calculated data for most published functions<br />
are those for the complexes used <strong>in</strong> the derivation of the functions themselves.<br />
Very promis<strong>in</strong>g results of rank order<strong>in</strong>g have also been obta<strong>in</strong>ed with the<br />
knowledge-based functions DrugScore 93 and PMF. 86,88,<strong>18</strong>2<br />
The task of rank order<strong>in</strong>g small (ca. 10–100) sets of related ligands with<br />
respect to a target can also be accomplished with methods that are computationally<br />
more demand<strong>in</strong>g than simple scor<strong>in</strong>g functions. The most generally<br />
applicable methods are probably force field scores augmented with electrostatic<br />
desolvation and surface area terms. An example is the MM–PBSA<br />
method that comb<strong>in</strong>es Poisson–Boltzmann electrostatics with AMBER MD<br />
calculations. <strong>18</strong>3 This method has been applied to an <strong>in</strong>creas<strong>in</strong>g number of<br />
studies, and it has led to promis<strong>in</strong>g results. 106–108,<strong>18</strong>4 Poisson–Boltzmann<br />
calculations have been performed on a variety of targets with many related<br />
computational protocols. 102,138,139,<strong>18</strong>5–<strong>18</strong>8 Alternatively, extended l<strong>in</strong>ear<br />
response protocols 112 can be used. The OWFEG grid method by Pearlman<br />
has also shown promis<strong>in</strong>g results. 114<br />
APPLICATION OF SCORING FUNCTIONS IN<br />
VIRTUAL SCREENING<br />
In recent years, virtual screen<strong>in</strong>g of large databases has emerged as the<br />
central application of scor<strong>in</strong>g functions. In the follow<strong>in</strong>g sections, we describe<br />
special requirements that scor<strong>in</strong>g functions must fulfill for successful virtual<br />
screen<strong>in</strong>g, and we <strong>in</strong>dicate the level of accuracy that can nowadays be<br />
expected from virtual screen<strong>in</strong>g.<br />
As discussed <strong>in</strong> the <strong>in</strong>troductory sections, the goal of virtual screen<strong>in</strong>g is<br />
to use computational tools together with the known 3D structure of the target<br />
to select a subset of compounds from chemical libraries for synthesis and