More than 33 million people live with HIV worldwide and 2.5 million people became infected with the virus in 2007. More than two million people died of an AIDS-related illnesses last year, according to the Joint United Nations Programme on HIV/AIDS.
The problem with treating the disease is that patients can build resistance to the drugs they are given. The HIV virus can mutate and change in different patients and this can make issuing the right combination of drugs difficult.
In response, staff at UCL will be using the combined supercomputing power of the UK and US 'national grids' to simulate a patient's biological responses to drugs. Doctors currently work by prescribing a course of drugs and then test whether these are working. One of the goals of the project is for such 'trial and error' methods to eventually be replaced by patient-specific treatments tailored to a person's unique genotype.
"To be able to tailor medical treatment to a person and their ailments, instead of giving them some average course of treatment - we're only going to get to that level of patient specificity if we use computational science and high performance computing, of that there can be no doubt," said Peter Coveney of the UCL Department of Chemistry.
The study will involve a sequence of simulation steps, performed across several supercomputers on the UK's National Grid Service and the US TeraGrid, using computational power roughly equivalent to that needed to perform a long-range weather forecast.
"The method is an early example of what is called the Virtual Physiological Human (VPH). The idea behind the VPH is to link networks of computers across the world to simulate the internal workings of the human body," he said.
The VPH - mainly a research initiative at present - allows scientists to simulate the effects of a drug and see what is happening at the organ, tissue, cell and molecular level.
"This study represents a first step towards the ultimate goal of 'on-demand' medical computing, where doctors could one day 'borrow' supercomputing time from the national grid to make critical decisions on life-saving treatments," said Professor Coveney
Data will be processed using four national grid service computers. Each computer facility comprises 64 dual CPU Intel 3.06GHz nodes and 4TB storage, while each data facility is configured with 20 dual CPU Intel 3.06GHz nodes and 18TB Fibre SAN storage, giving a total 44TB storage across all nodes available to users of the NGS.
JANET Lightpath, the high speed academic network, will provide end-to-end network capacity. The network can provide speeds within a range of a few tens of Mbit/s to a full 10Gbit/s, according to user requirements. Lightpaths can also be extended internationally by connection via GÉANT, the multi-gigabit pan-European education network, to reach other national research and education networks.
Lightpath will also have dynamic provisioning, where a user (or even an application), given suitable authorisation, can request a link to be established on demand and then released once the application has finished with it.
The university will also use the Texas Advanced Computing Centre Lonestar supercomputer, at the University of Texas. The supercomputer is a Dell Linux cluster, which comprises Dell PowerEdge 1955 blade servers and possess a peak performance of more than 55 teraflops.
"We have some difficult questions ahead of us, such as how much of our computing resources could be devoted to helping patients and at what price. At present, such simulations - requiring a substantial amount of computing power - might prove costly for the National Health Service, but technological advances and those in the economics of computing would bring costs down."
For the moment, Professor Coveney's group is continuing to look at all the protease inhibitors in a similar way. The VPH initiative will soon be under way with E72m of initial funding from the EU.
Coveney hopes it will boost collaboration between clinicians and scientists to explore the scope for patient-specific medical treatments based on modern modelling and simulation methods, and ultimately save lives.