366

27

Drug Discovery

Table 27.1 Stages of gene expression and their control

Stage

Description

Control (examples)

G

Genomeright arrow→transcriptome

(transcription)

Epigenetic regulation (networks)

T

Transcriptomeright arrow→proteome

(translation)

Post-translational modification

P

Proteomeright arrow→dynamic system

Distributed control networks

D

Dynamic systemright arrow→phenotype

Hormones

M

Metabolism

Allostery

Whereas traditionally drugs were sought that bound to enzymes, blocking their

activity, bioinformatics-driven drug discovery focuses on control points, at which

intervention using drugs can take place very effectively, as summarized in Table 27.1.

The results of expression experiments are thus carefully scrutinized in order to iden-

tify possible control points. Once a gene or set of genes have been found to be

associated with a disease, they can be cloned into cells and the encoded protein or

proteins can be investigated in more detail as drug targets (functional cloning).

The proteome varies between tissues, and different structural forms of a protein

can be made by a given gene depending on cellular context and the impact of the

environment on that cell. From the viewpoint of drug discovery, there are further

crucial levels of detail that need to be considered, namely the way that proteins are

subdivided structurally into discrete domains and how these domains contain small

cavities (active sites) that are considered to be the “true” targets for small-molecule

drugs.

Clustering as well as other pattern recognition techniques (Sect. 13.2) can be used

to identify control points in regulatory networks from proteomics and metabolomics

data. DNA, RNA, and proteins are thus the significant biological entities with respect

to drug development. The stages of drug development are summarized in Table 27.2.

Great effort is put into short-cutting this lengthy (and very expensive) process using

computational tools. For example, structural genomics can be used to predict, from

the corresponding gene sequence, the three-dimensional structure of a protein sus-

pected to be positioned at a control point. It may also be possible to compare active

sites or “specificity pockets” (these regions are typically highly conserved). Pharma-

cogenomics refers to the genotyping of patients in an attempt to correlate genotype

and response to a drug.

Another approach to target discovery is to automatically trawl through the entire

scientific literature—whatever is available on the web, including data that has not

been published in conventional journals, and even patient discussions on social

media—in order to get clues about what targets are associated with particular dis-

eases and what drugs are effective—or not—against those diseases, and which ones

might interact with identified targets. This is sometimes called “network-driven drug

discovery”.