BioSciences Group, Fujitsu

Docking studies have taken an increasingly important place in the workflow of structure based drug design (SBDD). Accurate docking of ligands into identified pockets adds a significant 3D descriptor to structure activity relationship (SAR)-based high throughput screening methods. Unfortunately, docking of even a modest-sized library can be computationally challenging, forcing many researchers to utilize computationally “cheaper” 2D screening methods, saving docking studies for only a subset of a library.

However, such methods may miss novel and unexpected interactions with active site pockets, thus decreasing the potential pool of leads. In an attempt to fully realize the potential of docking for library parsing, scientists at Fujitsu have harnessed the massively parallel computational architecture of the BioServer™ to support high throughput docking (HTD). As a proof of principal, the NCI screen set of compounds (4436 compounds) was loaded into Project Leader of BioMedCAChe™ connected to a deskside BioServer (128 processors). The entire compound set was docked (flex ligand versus flex active site) against a modeled active site of a drug resistance transporter expressed in chemi-resistant cancer cells utilizing FastDock, a proprietary PMF-based docking method developed by Fujitsu, running on the BioServer. Docking of the set took 8 hours. (Figure 1).In contrast a standard desktop PC (Dual Xeon with 1 GB of RAM) took over 16 days to complete these analyses.

Docking of NCI Screening Case Study

Performance result

Ian G. Welsford, Ph.D. Fujitsu Computer Systems, BioSciences Group