Machine learning & deep learning tools in pharmaceutical sciences: A comprehensive review

Abstract

Drug discovery and development is an important area of research for pharmaceutical industries and medicinal chemists. This classical approach demanded significant investments of time and resources to bring a single drug to market. Furthermore, the complexity and vast scale of data from genomics, proteomics, microarrays, and clinical trials present significant challenges in the drug discovery pipeline. Nevertheless, bioinformatics, pharmacoinformatics, and cheminformatics technologies have been developed thanks to breakthroughs in computational methodologies and a surge in multi-omics data, drastically shortening the time it takes to create new drugs. Large amounts of biological data stored in global databases are the building blocks for machine learning and deep learning methods. They make it easier to find patterns and models that can help find therapeutically active molecules with less time, work, and money. Machine learning and deep learning technology are vital in drug design and development. We have applied these algorithms to various drug discovery processes such as protein structure prediction, toxicity prediction, oral bioavailability prediction, de novo design of new chemical scaffolds, structure-based and ligand-based virtual screening, pharmacophore modeling, quantitative structure-activity relationship, drug repositioning, and clinical trial design. Historical evidence underscores the successful implementation of AI and deep learning in this domain. Finally, we highlight some successful machine learning or deep learning-based models employed in the drug design and development pipeline. Furthermore, there has been a notable increase in interest regarding the application of AI technology in hospital pharmacy settings, which has been discussed in this review. This review will be invaluable to medicinal and computational chemists seeking DL tools for drug discovery projects and hospital pharmacies.