Drug molecules are usually designed to bind to and modulate the action of a specific biological target. That such selectivity is rarely achieved in practice is demonstrated by the various side effects and the efficacy of drugs. Moreover, although some more recently developed drugs may have originally been designed for specificity, they are known to act through multiple targets in a way that is therapeutically essential.

Now, Brian Shoichet from the University of California, San Francisco, Bryan Roth from the University of North Carolina and co-workers have turned this to their advantage1. Instead of comparing similarities — such as protein sequence — between targets they have created a map that relates targets by the drugs that bind to them. They took 3,665 known FDA-approved and investigational drugs, and identified several previously unanticipated drug–target associations, including five that were shown experimentally to be potent (with binding constants <100 nm).

The interest in these newly acquired targets is threefold. Firstly, some off-target associations may be the major cause of side-effects that now warrant further investigation and, secondly, others may provide important new opportunities for as yet undiscovered therapeutic intervention. Finally, some of these drugs may exert their effect by acting at a different target to that previously thought and thus improve our knowledge of their mechanism of action.