Chemistry Reference
In-Depth Information
drug-like filters such as Lipinski's rule of five, rotable bond counts and solubility.
Constraints/filters also include structural ligand-centered binding-site, binding-
mode selection, and applicability domain in QSAR and excluded volumes in
pharmacophore methods. Combination approaches involves methods that
complement each other in screening speed as well as hit selection strategies to
yield enhanced performance. The faster method can be used first.
We can also make a division of the three-dimensional (3D) virtual screening
methods into three main groups with corresponding sub-groups as follows. Group
1 is 3D-LBVS, group 2 is SBVS and group 3 involves other 3DVS approaches. In
the first group, we have the following: a) 3d fingerprint-based screening
(alignment/comparison of molecular fingerprints); b) molecular fields comparison
(superposition/comparison of molecular fields); c) ligand shape matching
(superimposition/comparison of molecular surfaces); d) 3D-pharmacophore
matching (screening for a ligand conformation matching pharmacophore); e) 3d-
QSAR (information derived from the ligands conformational space).
Group 2 is SBVS and involves protein-ligand docking (exploration of the ligand
conformations space); binding site similarity (comparison of simplified
representation of macromolecular binding sites); molecular dynamics (whole
target structure and ligand atomic-level properties).
Group 3 involves some of the other 3DVS approaches. It is subdivided in
fragment-based VS (identification of small chemical fragments; chemogenomic
(identification of novel drugs and drugs targets); knowledge-based drug discovery
(data modeling and knowledge extraction).
Virtual screening methods have adjusted during the decade to screen, with
promising performance, increasingly larger chemical libraries involving even
more than millions of compounds. The continuous development of library
generation tools requires still further improvement of virtual screening methods to
meet the complex screening challenges. Effectively, further improvements of
individual VS methods include consensus-modeling strategies, integration of VS
and HTS, mining of actives and inactive from libraries, addition of off-target
identification, side effects prediction, improving efficiency and productive
