Korea@Home : the PC Grid

PC grid computing is based on the idea that the unused computational power of desktop PCs can be harnessed to create an aggregate computational resource greater than the largest clusters and the most powerful supercomputers available - at a fraction of the cost. PC grids are appealing because of their ability to provide enterprises with more than enough computing power to solve previously unsolvable problems, while at the same time offering a full return on their existing IT investments. Organizations choose PC grids because they offer distinct advantages over alternative solutions.

PCs are more powerful than ever before. Remarkably, GartnerGroup research tells us over 95% of today's PC power is wasted; between mouse-clicks, keystrokes, and spikes of routine application program activity, the average PC sits idle. Distributed computing technology provides the means to harness these valuable unused CPU cycles to create an aggregate computational resource known as a PC grid. Computationally intensive calculations that might otherwise take months to solve on a cluster or supercomputer can now be solved in hours or days on a PC grid. Even more exciting, PC grids open doors to new computing challenges never before considered possible.

Application to the Virtual Screening

Fig 1. Schematic Diagram of Korea@HOME

Korea@Home is a distributed computing project which searchs new drug candidates. We use novel computational methods and large scale distributed computing, to simulate timescales thousands to millions of times longer than previously acheived. This has allowed us to screen chemicals, and now to direct our approach to drug targets.

The docking and in silico ligand screening procedures can select small sets of likely lead candidates from large libraries of either commercially or synthetically available compounds; however, the vast number of such molecules make the potential size of this task enormous. To accelerate the discovery of drugs to inhibit several targets, we have exploited massively distributed computing to screen compound libraries virtually. The Korea@Home project was launched in Feb. 2002, and one year later, more than 1200 PC's have been recruited. This has created a 31-gigaflop machine that has already provided more than 1400 hours of CPU time. It has allowed databases of millions of compounds to be screened against protein targets in a matter of days. Now, the virtual screening software suitable for distributed environments is developed by BMD. It has been evaluated in terms of the accuracy of the scoring function and the search algorithm for the correct binding mode.