Introduction: Drug discovery is the process of designing new candidate medications for the treatment of diseases. Over many years, drugs have been identified serendipitously. Nowadays, chemoinformatics has emerged as a great ally, helping to rationalize drug discovery. In this sense, quantitative structure-activity relationships (QSAR) models have become complementary tools, permitting the efficient virtual screening for a diverse number of pharmacological profiles. Despite the applications of current QSAR models in the search for new drug candidates, many aspects remain unresolved. To date, classical QSAR models are able to predict only one type of biological effect (activity, toxicity, etc.) against only one type of generic target.
Areas covered: The present review discusses innovative and evolved QSAR models, which are focused on multitasking quantitative structure-biological effect relationships (mtk-QSBER). Such models can integrate multiple kinds of chemical and biological data, allowing the simultaneous prediction of pharmacological activities, toxicities and/or other safety profiles.
Expert opinion: The authors strongly believe, given the potential of mtk-QSBER models to simultaneously predict the dissimilar biological effects of chemicals, that they have much value as in silico tools for drug discovery. Indeed, these models can speed up the search for efficacious drugs in a number of areas, including fragment-based drug discovery and drug repurposing.
Keywords: CHEMBL; consensus predictions; fragments; moving average approach; multitasking quantitative structure–biological effect relationships models.