The VTS system requires extensive work initially to prepare the virtual protein structures obtained from the Protein Data Bank (www.rcsb.org). Dr. Dan Santiago and Dr. Yuri Pevzner developed a VTS prototype with ~1,500 proteins with which we were able to perform proof-of-concept testing, such as the publications (below) on kinases and natural products. Our VTS system uses a unique method to identify potential MOI/protein interactions of significance. One could simply compare docking scores of the MOI to a known inhibitor but many protein structures do not have any known inhibitors or potent inhibitors bound.  This presents a problem in consistency for interpreting MOI/protein interactions across the collection. Our solution is to use a library of small drug-like molecules from the NCI to pre-calibrate each protein to determine average scores. We use the NCI diversity set I of approximately 2,000 entries that represent the diversity of the NCI’s much larger collection of ~250,000 drug-like molecules. We use Schrödinger’s GLIDE docking application to perform the dockings and then we determine the average of the top ranking 200 and top ranking 20 NCI compounds.  These averages represent roughly the top ranking 5%, and 0.5% of the NCI compounds, respectively. We refer to this as calibrating the protein. When an MOI is run through VTS, its docking score against a protein can be compared to these averages for the protein to score the MOI/protein interaction. The calibration step with the NCI compounds provides a common reference for all proteins rather than having a different reference for each protein (e.g. a known inhibitor). 

VTS was originally conceived Dr. Wesley Brooks. Currently the VTS system is being scaled up to a production version by Alan Carregal and Rainer Metcalf who have developed a computer cluster for parallel processing. They are automating the protein preparation, calibration and the main VTS program to make use of their high performance computing environment. The objective is to develop a production version of VTS that can screen MOIs against 20,000+ proteins running in parallel.
 
Publications:
1.Santiago D.N., Pevzner Y., Durand A.A., Tran M.P., Scheerer R.R., Daniel K., Sung S.S., Woodcock H.L., Guida W.C., Brooks W.H. ‘Virtual target screening: validation using kinase inhibitors’ J. Chem. Info. Modeling (2012) 52:2192–2203.
 2.  Pevzner Y., Santiago D.N.,  von Salm J.F., Metcalf R.S., Daniel K.G., Calcul L., Woodcock H.L., Baker B.J.,  Guida W.C., Brooks W.H. ’ Virtual target screening to rapidly identify potential  protein targets of natural products in drug discovery’ AIMS Mol. Sci. (2014) 1:81-98.