Virtual high-throughput screening of molecular databases

Abstract

Virtual high-throughput screening (vHTS) is an efficient and widely applicable method used to identify initial hit compounds for pharmaceutical research. Despite its widespread use, several aspects of protein structure-based vHTS can still be optimized, particularly its accuracy and speed in generating results. Recent developments that address these issues include machine learning and implicit solvation methods. Various machine learning methods are available to improve vHTS accuracy, for example, target-specific optimization of scoring functions, the integration of essential protein-ligand interactions, and the application of negative training data. Implicit solvation methods are exemplified by the molecular mechanics Poisson-Boltzmann solvent accessible surface area approach. Furthermore, grid computing and intelligent database screening approaches are used to improve the speed of vHTS. © The Thomson Corporation.