Chemistry Reference
In-Depth Information
protein a number of compounds that inhibits and leads to the highest probability of being a good drug candidate
[38].
Virtual screening can either be performed by the docking of potentials ligands into a protein binding site or by a
similarity-based method. A similarity-based approach can be performed using a pharmacophore searching
procedure, where the distances between GRID MIFs derived from the protein structure binding site can be used to
drive a search for molecules with desired pharmacophore profile. In this way, the distance-based description of the
protein is compared to the distances between structural properties in the potential ligand. This procedure could be
taken as a preliminary task, enabling the analysis of large datasets, of which interesting hits should be further
analyzed by more accurate methods [39].
Virtual screening tools grow with the knowledge available for a particular drug target and pharmacophore data. New
developments indicate that progress can be made by combining pharmacophore filtering, docking methods and using
several scoring functions in parallel [40]. Many methodologies are used for analyses within 3D constraints of an
active target site. However, pharmacophore approaches have proven to be one of the widely applied methods in
virtual screening [39].
Ahlstrm
et al
. [39] studied a set of strategies for structure-based design using GRID molecular interaction fields
(MIFs) to derive a pharmacophoric representation of a protein. Thrombin was chosen as the model system. A
pharmacophore representation of the thrombin protein based on GRID MIFs was used for the virtual screening
methodology. A new procedure to perform a bioisosteric substitution of one scaffold for another, called scaffold
hopping was used in this study. The starting point of the procedure was the selection of a template scaffold, which
could be used during the search. When the scaffold is chosen, the number and positions of anchor points (atomic
positions where the scaffold can be chemically modified) is decided.
Neves
et al
. [41] studied the development of a new ligand-based strategy combining important pharmacophoric and
structural features according to the postulated aromatase binding mode, useful for the virtual screening of new
potent non-steroid inhibitors. A small subset of promising drug candidates was identified from the large NCI
database. The screening was based on the common features of a training set of second and third generation
aromatase inhibitors combined with heme coordinating fragments inclusion volumes, and several druglikeness
filters.
ADME PREDICTION
There are many methods and technologies that have been improved in the last years and are currently used to help in
the drug discovery process. Many potent lead compounds for particular drug targets have been discovered;
nevertheless the number of drugs being introduced in the market is much smaller. Most lead compounds fail in pre-
clinical assays, because they present inadequate ADME properties (absorption, distribution, metabolism and
excretion) and/or are toxic to humans [42].
It is now recognized that the definition of a successful drug is an appropriate balance of potency, safety and
favorable pharmacokinetics. In order to reduce time, costs and the distance between the discovery of a new lead and
the clinical approval, new computational techniques have being developed with the intent to predict pharmacokinetic
properties and select compounds with suitable pharmacokinetic profiles, helping in guiding the drug discovery
process and evaluation of drug candidates. [42-44].
In parallel with the experimental ADME field of research,
in silico
methods have been developed for specific
in
vitro
assays or specific pathways in systems. These computational models are built based on experimental data. It is
clear that a combination of
in silico
methods with selected
in vitro
tests can provide a cost-effective complement to
exhaustive
in vitro
screening and animal experiments. [28].
The molecular interaction fields are useful tools employed computationally which can lead to models for studying
ADME parameters. The interaction of a molecule with a hydrophobic probe can be used to highlight the
