BHSAI

DOVIS (DOcking-based VIrtual Screening) is a tool for virtual screening of chemical databases containing up to millions of small, drug-like compounds. The designed docking-based virtual screening pipeline uses the AutoDock 4.0 program[1] as its docking engine and is integrated into an HPC environment. It removes many technical and administrative complexities involved in employing AutoDock for large scale virtual screening.

The DOVIS HPC application allows end users to quickly identify potential bioactive compounds, within large databases, for a given protein target. The DOVIS pipeline:

• Accepts ligands in sd format and proteins in either pdb or mol2 format. The docked ligands are saved in sd files embedded with AutoDock4 score.
• Has an approximate throughput of ~700 compounds/CPU/day. [2,3]
• Has an enhanced parallelization scheme with automatic restart function, allowing large chemical databases to be screened in one session.
• Provides job control functions to allow users to start, monitor, stop and restart jobs.
• Provides an interface to third-party scoring programs to rescore docked ligands and a capability to cluster rescored ligands.
• Is easily accessible through its Web-based graphic user interface (GUI) using the User Interface Toolkit (UIT).
• Is currently installed at the Army Research Laboratory Major Shared Resource Center on the jvn and mjm clusters; the application can be ported to any Linux-based cluster system.

The following screenshot of the DOVIS GUI shows the table of docked ligands and the visualization of the structure of a selected docked ligand with its protein target:

• Ensemble docking of ligands to multiple protein conformations
• Post-docking optimization of ligands bound to a protein