Integration of virtual and high-throughput screening

Key Points High-throughput (HTS) and virtual screening (VS) have progressed rather independently over the years. However, these disciplines have similar goals and are highly complementary. There are good indications that drug discovery research will increasingly benefit from an integrated approach to screening. A diverse array of VS methods has been developed, including structural queries, pharmacophores, molecular fingerprints, QSAR models, diverse cluster analysis tools, statistical techniques and docking calculations. In addition, VS techniques have been implemented to filter large databases for compounds with desired or undesired chemical groups, drug-like character, preferred solubility and absorption characteristics, or oral bioavailability. Both small-molecule- and structure-based VS have recently produced several success stories in the search for novel inhibitors or antagonists of diverse biological targets. Some VS methods have been introduced or adapted for the analysis of HTS data, taking into account that such data sets are usually noisy and error prone. Prominent among these methods are different partitioning and clustering algorithms that can derive predictive models of biological activity from screening data. Similar approaches are used to interface HTS and VS directly. At present, this is best accomplished by the application of iterative screening strategies, such as focused or sequential screening. Although the details of such strategies can differ considerably, they have in common that small subsets of compounds are computationally selected from large databases and assayed. On the basis of the obtained results, the search for biologically active molecules is further refined in subsequent iterations. In several case studies, sequential screening has yielded significant improvements in hit rates over random screening. It is not uncommon for iterative screening to achieve hit rates between 10% and 40% (by markedly reducing the number of compounds to be tested). As the size of compound databases and the number of available screening targets rapidly increase, it is conceivable that combined computational and biological screening might soon become a focal point of pharmaceutical research, despite the advances that are being made in the HTS arena towards even higher throughput. High-throughput and virtual screening are important components of modern drug discovery research. Typically, these screening technologies are considered distinct approach