Guohua Huang*, Fengxia Yan and Duoduo Tan Pages 562 - 572 ( 11 )
Drug discovery and development is not only a time-consuming and labor-intensive process but also full of risk. Identifying targets of small molecules helps evaluate safety of drugs and find new therapeutic applications. The biotechnology measures a wide variety of properties related to drug and targets from different perspectives, thus generating a large body of data. This undoubtedly provides a solid foundation to explore relationships between drugs and targets. A large number of computational techniques have recently been developed for drug target prediction. In this paper, we summarize these computational methods and classify them into structure-based, molecular activity-based, side-effectbased and multi-omics-based predictions according to the used data for inference. The multi-omicsbased methods are further grouped into two types: classifier-based and network-based predictions. Furthermore， the advantages and limitations of each type of methods are discussed. Finally, we point out the future directions of computational predictions for drug targets.
Drug targets, off-target, side-effect, machine learning, heterogeneous network, gene expression profile.
Provincial Key Laboratory of Informational Service for Rural Area of Southwestern Hunan, Shaoyang University, Shaoyang, Hunan 422000, College of Science, National University Defense Technology, Changsha, Hunan 410073, School of Finance and Statistics, Hunan University, Changsha, Hunan 410081