Current advances in ligand‐based target prediction

Target identification for bioactive molecules augments modern drug discovery efforts in a range of applications, from the elaboration of mode‐of‐action of drugs to the drug repurposing to even the knowledge of side‐effects and further optimization. However, the traditional labor‐intensive and time‐consuming experiment methods obstructed the development. Driven by massive bioactivity data deposited in chemogenomic databases, computational alternatives have been proposed and widely developed to expedite the validation process. By screening a compound against a protein database, it is possible to identify potential target candidates that fit with this specific compound for subsequent experimental validation. In particular, ligand‐based target prediction methods have made tremendous progress in the past decade due to their flexibility, relatively low computational cost, and remarkable predictive performance, and are still moving forward. In this review, we present a comprehensive overview of ligand‐based target prediction methods including similarity searching, machine learning and algorithm stacking, and the strategies to validate these methods. We also discuss the strength and weakness of the existing data sources for model development and outline the challenges and prospects of ligand‐based target prediction. It is expected that the topic discussed in this review should guide the development and application of ligand‐based target prediction and be of interest to the audiences for wider scientific community. This article is categorized under: Data Science > Chemoinformatics