‘blockbuster’ drug with annual sales in excess of $1bn. However, the risk with such a
drug programme is that the concept underlying its action will not previously have
been confirmed. Alternatively, the drug may be in an existing class in which case it
will have competitors and the aim will be to make it ‘best in class’.18.2.3 Selection of a screening target and assay procedure
In silicopharmacology
It has been estimated that in the process of discovering and developing the 800 drugs
that are currently licensed for clinical use, pharmaceutical companies have synthe-
sised and evaluated against their particular narrow target field over 10 million
compounds. Individual pharmaceutical companies currently hold chemical ‘libraries’
of up to 2 or 3 million compounds. Such a large resource is available for screening
against new therapeutic targets. Moreover, in the last decade huge advances have
been made in compiling computer databases of both the structural, physical and
biological properties of existing drugs and structural classification of all known
protein targets. These databases are highly sophisticated. For example, the ligand
database will contain such information as molecular weight, molecular structure,
H-bond donor numbers, polarity, surface area, lipophilicity (expressed as a logP
value, wherePis a measure of the partition coefficient between water and an organic
solvent usually 1-octanol), three-dimensional conformation, spectral properties and
known biological targets (receptors, enzymes), quantitative structure–activity rela-
tionships (QSARs) and structure–absorption, distribution, metabolism and excretion
relationships. Similarly, the protein database will contain all known information
about primary, secondary, tertiary and quaternary structure, binding site details of
the protein and structural details of known ligands. Knowledge of the amino acid
sequence of a target receptor enables a computer model of the structure of the agonist-
binding site to be made.
The availability of such data sets has given rise to the interdisciplinary field
of chemogenomicswhich aims to predict gene/protein/ligand relationships and
hence enable predictions to be made of potential ligands for new targets long before
any screening has taken place. Such predictions will be based on the assumptions
that chemically similar compounds should share common targets and targets
sharing similar ligands should share similarities in their binding sites. This use of
computer-based data to aid drug discovery is known under the names ofin silico
pharmacology,computational pharmacologyandcomputational therapeutics. Such
computer-based predictions are used to identify potential drug targets and to test the
models. They are run in parallel with high-throughput screening techniques in drug
discovery programmes.High-throughput screening
The screening of large chemical libraries for a specific type of biological activity is
generally based either on binding activity of the test compound against a selected
target preparation or the ability of the compound to activate or inhibit some particular725 18.2 Drug discovery