Advanced Research Computing


HIV research using UCL computing platforms has a real-life impact

Buying into part of the Legion HPC cluster has helped Professor Andrew Phillips to enhance HIV research by deploying computer simulation models of HIV infection, progression, resistance and treatment

Andrew's co-researcher Valentina Cambiano with colleagues from the Zimbabwe Ministry of Health at a clinic

16 September 2015


Research into the evolution of HIV and its progression as a disease has often had to rely on retrospective observation of clinical and non-clinical variables, and unverified assumptions. The time-lag between events and understanding of them is an integral part of the disease's story.

The issue is not of just scientific knowledge within the study of HIV itself, but also includes human behavioural factors. The key is to have a deeper understanding of existing issues, and the options available amid the creation of policy by government and non-governmental organisations, across the world.

The possibilities of new techniques in computational science and their applications look set to become the foundations of a new armoury of predictive models within HIV research. Advanced HIV modelling allows new scenarios to be envisaged, rather than just using past events as a basis for prediction. Consideration of drug resistance, treatment recipient compliance, the effects of earlier-timed drug intervention amid viral load considerations, personal behaviours, and self-treatment, are the key themes of this new level of forensic study. These new models, using the power of high performance computing resources such as UCL's Legion cluster, have great potential to enhance data and the delivery of care.

This work has the potential to influence policy in the UK and developing world - including informing World Health Organisation (WHO) recommendations and Department for International Development (DfID) expertise.

What we did

Professor Andrew Phillips
Professor Andrew Phillips, a senior academic in the UCL Department of Infection and Population Health, based at the Royal Free Hospital, purchased a dedicated section of UCL's Legion cluster to carry out this mathematical modelling. Purchasing a section of the Legion system allows researchers to get better value for money than buying the equivalent hardware as a stand-alone system because users can take advantage of the specialist maintenance and support offered by the UCL Research Computing team.


Professor Phillips and his colleagues' use of Legion nodes has also benefited by their being able to use more computational resources than they purchased because the computational work can migrate seamlessly from the licensed section of Legion, into the main UCL-wide system when necessary. This allows even greater use of computer resources to further accelerate simulations.

Professor Phillips says:

" We have created a range of computational models reconstructing aspects of the HIV epidemic and its underlying factors. By better understanding the driving forces behind the trends and events that we have seen in the past, we can make more informed choices to reduce HIV infections, and to improve treatment outcomes. The use of the Legion computer system has been fundamental to achieving this knowledge and its impact, and I would say that my HIV research would not have happened without this advanced IT technology.


Evidence of Andrew's (and co-authors) world-class computerised simulation work can be seen in numerous research outputs including key published and in-press papers. To date, 12 published papers in HIV research have cited the Legion cluster. Specific real-world outcomes include:

  • Prediction of cost-effectiveness of potential policies to adopt in response to different levels of pre-treatment HIV drug resistance - the modelling will help inform WHO recommendations on monitoring of HIV drug resistance in people starting antiretroviral therapy at an early date.
  • Viral load-informed differentiated care research, demonstrates that there are inefficiencies in the current approach to monitoring of patients on antiretroviral therapy in sub-Saharan Africa and that it is critical for the sustainability of treatment programmes to transition to a more effective and cost-efficient approach to monitoring people on these therapies.
  • Projected life expectancy of people with HIV according to timing of diagnosis showed that if low rates of virologic failure observed in treated patients continue, predicted life expectancy is relatively high in people with HIV who can access a wide range of antiretrovirals.
  • In an analysis of suggested HIV self-testing in low income countries, the modelling (for Zimbabwe) revealed that self-testing could lead to substantial health care savings.

Direct HIV expertise stemming from enhanced computational science is visible in the UK and overseas HIV knowledge base. The Legion system has not just handled data, it has played a part in its creation and realisation as an outcome.

The beneficial effect of earlier intervention of antiretroviral drugs within UK policy compared to existing parameters, has been framed and evidenced; wider insights into HIV drug resistance, infection progression in sub-Saharan Africa and the cost/health benefit ratio of self-treatment in such locations are now also better understood - with subsequent gains to global HIV policy.

Real advances in computer simulation architecture and resulting quantitative and predictive accuracy have been gained: computational science can address major research issues in analysis and epidemiology and, generate a new stream of scientific leaders.