Polypharmacological Drug-target Inference for Chemogenomics

Pharmacological drug actions are often caused by multi-target effects. While most of the currently approved synthetic drugs were designed to interact with a single 'on-target', these chemical agents often interact with additional 'off-targets'. Understanding and rationalizing these multiple interactions will be indispensable for the design of future precision medicines. We employed computational predictions of drug-target interactions to analyze functional drug-drug relationships. 900 approved drugs were represented in terms of their predicted activity fingerprints, considering 1158 potential target activities. A drug relationship network was constructed based on fingerprint similarity. The resulting network graph highlights clusters of compounds sharing similar predicted on- and off-targets, and allows to identify mutual targets of drugs that were originally developed for different therapeutic indications. Such an analysis offers straightforward access to spotting potential off-target liabilities and drug-drug interactions, as well as drug repurposing opportunities.