Supercomputing grids close ranks to meet medical challenge
Two supercomputing networks have successfully joined forces in a distributed simulation of the effectiveness of drugs on mutant strains of the Human Immunodeficiency Virus (HIV). Although the Distributed European Infrastructure for Supercomputing Application (DEISA) and the GridAustralia-Australian Partnership for Advanced Computing (APAC) infrastructures use incompatible underlying platforms (UNICORE 5 and Globus Tool Kit, respectively), the researchers were able to spread the computing tasks over the two high performance computing (HPC) grids. As HIV is highly mutable, it frequently becomes resistant to drugs that reduce patients' viral loads and bind and inhibit critical viral enzymes. As a result, patients will then have to change their drug regime. It is essential to select the right drug in order to provide the best possible treatment, as well as to prevent the development of further drug resistance. The simulations run by DEISA and APAC are intended to help this process by testing the effectiveness of particular antiviral drugs against a number of mutant HIV strains. They analyse average interaction energies between anti-HIV drugs and the HIV protease strain, hoping to thus provide an accurate assessment of the drugs' effectiveness. A huge number of calculations is needed for this, and this is why the process necessitates the processing power of supercomputers. Researchers were sure that time-to-solution would be further reduced by employing several supercomputers in a grid and spreading the tasks over different grids at the same time. This is the first time that a reliable, automated bidirectional data transfer between the European DEISA grid and the Australian APAC grid has been demonstrated.