Structure‐Based Drug Design

Abstract

Structure‐based drug design is achievable by exploiting the three‐dimensional information intrinsic in structural models of proteins that are involved in human disease. Many other tools (biochemistry, cellular and whole‐animal biology, organic synthesis, pharmaceutics, and computational analysis) must by tightly integrated to fulfill the promise of the structure‐based approach, and the use of this approach does not automatically guarantee success. However, the successes that have been achieved make it clear that structure‐based methods provide an efficient, effective, and elegant approach to medicinal chemistry. Examples are described in this chapter of the use of structure‐based methods in designing compounds that have entered human clinical trials and, in many cases, were launched as products. Some of the lessons that have been learned about best practices in the application of structure‐based methods are summarized.

The PDBbind Database: Methodologies and Updates

“…Structure-based drug design has been widely accepted as an established technology for drug discovery research in academia and the pharmaceutical industry. Several new drugs, whose invention was largely facilitated by structure-based drug design methods, have already reached the market …”

Application of 3D-QSAR in the Rational Design of Receptor Ligands and Enzyme Inhibitors

“…The rapid increase in available structural information, attributed to the advances in molecular biology to generate target proteins in adequate quantities, and the equally impressive gains in NMR and crystallography to resolve protein structures at atomic level, has stimulated the molecular-modeling community in the development of useful computational tools (e.g., docking and de novo design software), able to transform the knowledge of the topography of target binding sites into new drugs. Several successful applications of the so-called −structure-based drug design× (SBDD) approach have been recently reported, confirming its usefulness in medicinal chemistry [2]. On the other hand, the traditional −ligand-based drug design× (LBDD) paradigm is the most suitable approach to design new bioactive molecules when the target structure cannot be directly observed, or when its interaction with the drug is so complex that −hidden× structural features, which may not be revealed by the static topography of a protein ± drug complex, can influence…”

“…We prepared a series of N-substituted MLT derivatives, carrying substituents with different size and shape at the indole Natom. Among these compounds, N-benzylmelatonin was able to fit the MT 2 out-ofplane pocket and was predicted by the 3D-QSAR models to be a selective MT 2 antagonist. The pharmacological data confirmed this prediction, N-benzylmelatonin being endowed with a good affinity for the MT 2 receptor, and with no intrinsic activity.…”

Semi-equivariant conditional normalizing flows, with applications to target-aware molecule generation

“…Data consisting of sets of three-dimensional points, sometimes also referred to as point clouds, appear in a number of scientific and engineering settings. Examples include molecular chemistry (Beddell et al 1976, Lapatto et al 1989, Miller et al 1989, Davie et al 1991, Blundell 1996, Varghese 1999, Campbell 2000, Berman et al 2003, Hardy and Malikayil 2003, Congreve et al 2005, Zardecki et al 2016, Wu et al 2018, Jing et al 2020, in which the points are the atoms making up a molecule; high energy physics (Qu and Gouskos 2020, Kansal et al 2021, Pata et al 2021, Shlomi et al 2021, Duarte and Vlimant 2022, Qu et al 2022, in which the points represent locations of particles produced in a collider; and computer vision (Liu et al 2019b, Bello et al 2020, Guo et al 2020, in which the points are derived from the output of a three-dimensional sensor such as Lidar or Time-of-Flight. In many cases, these point sets can be effectively described by a 3D graph; that is a graph whose vertices correspond to points in R 3 .…”