Using filters in virtual screening: A comprehensive guide to minimize errors and maximize efficiency

Virtual screening (VS) is increasingly becoming key in our search for novel drugs against emerging therapeutic targets. Accordingly, this book chapter provides a brief description on the best practices adopted in the pre-VS process. It starts with an illustration of available ligand libraries, their types, and various parameters to be considered when preparing them for VS. This is followed by a detailed description of the best filtering practices in the field that ensure the highest quality in the resulting hits. Examples of these filters include various drug-like and lead-like rules, PAINS filters, and other promiscuity-related filters. Knowledge-based filters such as pharmacophores and other ligand-based approaches are also described within. Finally, this chapter also highlights the latest advances in machine learning and how they have been successfully employed to deal with huge ligand libraries to live up to the future challenges in the drug development arena.