Generative Drug Discovery
Quentin Vanhaelen, Alex Aliper, Alex A. Zhavoronkov · 2024
The advent of generative chemistry has provided novel opportunities for the de novo design of drug-like compounds. Over the last five years, progress has been made and several de novo designed compounds have entered stages of preclinical validation while some are undergoing clinical trials. The applications of AI are diverse, and the pharmaceutical sector envisions a future in which artificial intelligence is routinely used in drug discovery. In this chapter, we discuss the evolution of the field and review several technological breakthroughs that made these achievements possible. We discuss case studies to show how the combination of powerful deep learning techniques and novel architectures such as Transformer with the emergence of tools and metrics to evaluate generated molecular structures allows the design of compounds with suitable drug-like properties.