Despite the rapidly growing knowledge of functional and structural information regarding pharmaceutically relevant targets during the past decade, target-based drug discovery has remained a high-cost and low-yield process. Particularly, single-target drugs often turn out to be less
effective in treating complicated diseases such as cancers, metabolic disorders and CNS diseases. However, discovering compounds that are effective against multiple desired targets raises an enormous challenge to the current mode of drug innovation. Computational chemogenomics approaches aim
at predicting all potential interactions between small molecular ligands and biomolecular targets, thus the derived information can be directly applied to “design in” (i.e. engineer desirable binding spectrum) and “design out” (i.e. eliminate the unwanted interactions)
specific biological activities. The present review will focus on introducing the recent methodological development and successful applications of structure-based and ligand-based approaches on predicting the ligand binding profiles, which is the very first and essential step toward rationally
designing the multiple-target ligands. Structure-based methods (e.g. binding site mapping and inverse molecular docking) generally require the structures of known targets to navigate the receptor-ligand binding space, while ligand-based approaches (e.g. chemical similarity analysis and pharmacophore
search) can only rely on the series of active compounds to derive the structural characteristics for describing certain biological activities.