Feature Stories - Supercomputers target HIV
It is vital research. AIDS has killed more than 25 million people between 1981 - when it was first recognised - and 2005, making it one of the most destructive pandemics in recorded history. Medical science urgently needs to find new approaches to tackle the virus, but the research is very challenging. The 'Distributed European infrastructure for supercomputing applications' (DEISA) has helped researchers in developing molecular simulations of HIV mechanics. Over the course of five years and two projects, DEISA linked Europe's most powerful supercomputers via a network, and developed software that made it easy for researchers to access and use their massive processing power. They also developed support services to ensure researchers can get the greatest benefit from the available equipment. In the course of their work, they also set up the 'DEISA extreme computing initiative' (DECI) to support leading-edge scientific research in Europe; research that would benefit from the enormous computing power DEISA has made available. Researchers at the RNAHIV project, for example, used the infrastructure to seek a better understanding about how drug molecules bind to Ribonucleic acid. RNA is one of the molecules that form the basis for all life on the planet - along with Deoxyribonucleic acid (DNA) and proteins - and it directs the creation of proteins. But most anti-HIV drugs target viral proteins rather than viral RNA. As a result, they can fail because the virus develops drug resistance. Currently, there is still a lack of knowledge about how HIV RNA interacts with human cellular proteins to enhance the viral transcription, an important step in viral replication. Attempting to inhibit such interaction could lead to more promising anti-HIV drugs and have an important impact on the field of drug design. To tackle diseases such as HIV, you need to develop drugs that bind to a specific region of viral RNA, called the 'Trans-activating response' (TAR) region. But the problem is that standard computational approaches are just not very good at accurately predicting how or where drug molecules will attach to RNA. Normal drug design tools struggle to do this. 'But by focusing on the physics that governs the interactions that occur when molecules bind to each other, we were gaining good insights that might help in the development of RNA-based drugs,' says Paolo Carloni, coordinator of the RNAHIV project and professor of Computational Biophysics at the German Research School for Simulation Sciences, a joint graduate school of RWTH Aachen University and Forschungszentrum Jülich (FZJ), Germany. The RNAHIV project began in 2008 and was completed in 2010. It gathered researchers from around the world, including the SISSA/ISAS in Trieste Italy, ETH Zurich in Switzerland, the University of Washington in the USA, and the University of Ho Chi Minh City in Vietnam. Over the course of 24 months, the experts sought to simulate the dynamics of the binding mechanism between HIV and RNA. But studying these interactions is a big problem computationally. 'We were dealing with several thousand atoms and, in order to carry out simulations, you have to know where each goes and be able to follow its motion,' explains Dr Carloni. 'This is something that requires a lot of computer power!' These extremely complex simulations are very challenging, but Dr Carloni believes this method provides a much more rigorous description of the process by which the drug binds to the RNA. 'In conventional drug design, you end up only with a prediction of where the drug binds, but you don't learn anything about the way that the drug molecule travels to the RNA and latches onto it,' he notes. Exciting new approach Dr Carloni stresses that this is very useful research. 'It not only suggests a new way to design drugs, the methods we have developed can be applied to the study of any kind of reaction between proteins and DNA or between proteins and RNA. Those types of reactions take place in an enormous number of cell processes.' It was an exciting project, he stresses, because it really brings together physics and medicine to solve a very challenging problem. Their work involved a three-step process. The researchers began by drawing on their theoretical knowledge of biophysics to predict how the RNA and drug would interact, and they used spectroscopy techniques in an experimental phase to test whether their predictions were correct. The spectroscopy data was then used to inform the simulations of the molecular dynamics. The computing time provided by EU-funded DEISA played a vital role in the success of the RNAHIV project. Computations of the RNAHIV research project were performed on the Cray XT4/XT5 system at the IT Centre for Science in Finland. There were three independent runs, which used up to 256 CPU cores each. About 250,000 CPU hours were invested on the core of the production simulations and the hardware was complemented by software support. 'But DEISA has done more than just provide access to supercomputers. By supporting projects, such as RNAHIV, DEISA also opened up the use of supercomputers for other projects related to human health. And by making it possible for scientists from a developing country, like Vietnam, to participate, DEISA spread the benefits still further,' says Dr Carloni. 'Carrying out experimental research is difficult for our Vietnamese collaborators, but with computers they can work remotely. In our case, DEISA provided a wonderful opportunity for Vietnamese researchers to enter into the exciting field of RNA drug design, and to be able to take the knowledge back home.' RNAHIV was not the only HIV research project to benefit from DEISA's expertise and power. The EU-funded 'Virtual laboratory' (ViroLab) project also used DEISA to run computationally expensive molecular dynamics simulations. A big challenge with the HIV virus is the variety of strains associated with the disease. Some strains are more resistant to one drug, but more susceptible to another. ViroLab supports the decision-making process of doctors treating patients with HIV. The treating physician starts by analysing the HIV viral sequence infecting their patient. Once the analysis is complete, the doctors receive a selection of appropriate drugs. ViroLab automatically retrieves data from a variety of sources. It finds the resistance rules provided by the commonly used HIV drug resistance interpretation systems through drug ranking databases like Rega, HIVdb and ANRS. Meanwhile, anonymised patient data comes from participating hospitals, while the literature related to the particular strain of HIV is extracted from the US National Library of Medicine. As part of ViroLab's work, the project also developed and published a study validating their 'Binding affinity calculator' (BAC) tool, which helps determine which drugs would be more effective and which would be more resistant. By using DEISA-powered simulations, the ViroLab project acquired atomic-level insight into molecular-level interactions between a strain of HIV virus and particular drugs. This allowed the project to probe how resistance associated mutations interact with one another and cause changes in drug binding. This type of research could be used to assess the existing rules used to select drugs. But even more hopeful is the work's potential impact on future drug design. RNAHIV and ViroLab were just two projects in the life sciences that were helped by the supercomputing hardware, software, support and expertise provided by DEISA2. DEISA2 was funded to the tune of EUR 10.24 million (of EUR 18.65 million total project budget) under the EU's Seventh Framework Programme for research, 'e-Science grid infrastructures' sub-programme. Useful links: - 'Distributed European infrastructure for supercomputing applications' - DEISA2 project data record on CORDIS - e-Infrastructures programme / projects - RNAHIV project - ViroLab project Related articles: - Linking supercomputers to simulate the sun, the climate and the human body - Climate models run supercomputer catwalk - Supercomputing calms troubled waters - Supercomputing gets its own superhero - The grid: a new way of doing science - Europe's fusion researchers to tap into top supercomputing resources - ViroLab to use DEISA infrastructure