In silico drug discovery: a machine learning-driven systematic review

Abstract

This systematic review, which was carried out between 2018 and 2022 in accordance with PRISMA principles, assesses how machine learning (ML) and other computational approaches are integrated into drug discovery, with a focus on virtual screening (VS). The main goals are to evaluate the state of in silico drug-target interaction prediction techniques, gather useful computational tools, and provide model building help. The study emphasizes the significance of ML, molecular docking, bioinformatics, and cheminformatics in improving drug development efficiency by assessing 201 papers, of which 119 met inclusion criteria. It serves as a methodological guide for researchers, emphasizing on the effective use of computational approaches and decision-making improvements. This study relates computational techniques to drug development, discusses present constraints, and recommends future research topics with the goal of accelerating and improving therapeutic agent discovery. In summary, this systematic review highlighted numerous major tools, databases, and techniques that are critical in computational drug discovery, including the ChEMBL Database, Random Forest (RF) Algorithm, Extended Connectivity Fingerprints (ECFP), and RDKit. These tools and techniques highlight the transforming power of computational methods in pharmaceutical development. They offer researchers the ability to develop new computational models and improve drug development processes, thereby enabling the rapid advancement for new therapeutic agents via robust platforms.