1 Application of a Bioinformatics Approach to High-throughput Docking for Drug Discovery

Abstract—we have developed a high-throughput docking (HTD)-based virtual screening scheme, termed HiPCDock, for drug discovery and development. To improve the statistical significance of our screening results, a bioinformatics approach, motivated by sequence alignment package BLAST, was implemented so that we can estimate the confidence of our prediction accuracy. The model was validated by docking ten known thymidine kinase (TK) binders into the enzyme and the real inhibitors showed significant statistics of low probability and expectation values compared to those random compounds extracted from ChemBank. It was also demonstrated that HiPCDock was able to efficiently recover the ten known TK inhibitors from a dataset including 990 decoy compounds randomly selected from ChemBank. Our HiPCDock is currently implemented based on only one configuration template file for job submission, which makes it very easy to use for both computational experts and experimental scientists who have little molecular docking experience. With just one command line, users can submit massive parallel docking jobs for their screening of millions of compounds against a specific target of interest. The statistical model will be very useful to guide the decision-making for lead discovery and optimization. Thus HiPCDock is an elegant, professional integrated package for high-throughput molecular docking and drug discovery. We are currently developing a web-based user interface to further simplify the process. Index Terms—extreme value distribution, high-performance computing-based docking, virtual screenin