DOVIS: A Tool for High-Throughput Virtual Screening

We developed a Docking-based Virtual Screening (DOVIS) pipeline to predict how small molecules may interact with a given protein, so that we can rank a large database of molecules based on their predicted affinities to the protein. We used the program AutoDock as the docking engine and OpenBabel as the molecular data model. We designed and implemented a novel parallelization scheme using a file-based inter-process communication protocol to control parallel jobs. A Web- server/Web-page architecture was developed based on the User Interface Toolkit (UIT) to manage DOVIS jobs on high performance computing (HPC) platforms and provide a Web-page-based graphical user interface (GUI). The DOVIS pipeline is running on JVN at the Army Research laboratory (ARL) Major Shared Resource Center (MSRC). Scientists at several Department of Defense (DoD) labs are currently using the DOVIS pipeline in their research projects.