Super duper, our very own computer

When Premier John Brumby announced in San Diego in June 2008 that Victoria would get its own 400 teraflop supercomputer, he was more excited than this reporter has seen him before, and that’s saying something.

Brumby, who has a deep and abiding interest in life science research and biotechnology, had waited many years for this announcement. Victoria would host a supercomputer dedicated purely to the life sciences, and one that would rival the best in the world, he said.

“This is a big deal – it’s the biggest in Australia and the biggest for the life sciences in the world,” Brumby boasted. “It is 400 teraflops – to give you a comparison, the University of California, San Diego, theirs is 40.”

He made the announcement standing in front of said 40 teraflop computer at the San Diego Supercomputer Centre, his nonplussed American hosts looking on.

Not only will the supercomputer be fast and powerful, but it won’t be overly expensive. While $100 million sounds a lot, not all of it is for the computer itself.

The Victorian Life Sciences Computation Initiative (VLSCI) promises a lot of software and human expertise-building along with the hardware, and aims to accelerate research in areas such as cancer, cardio-vascular and neurological disease, chronic inflammatory disease, bone disease and diabetes. It is purely for the life sciences.

Half of the money is coming from the University of Melbourne and the other from the state government.

The money will be spent on establishing a Life Science Computation Centre (LSCC) at the Parkville campus of the university as well as a Peak Computing Facility (PCF), hosting the supercomputer, which will be available to scientists from the host institution as well as the other universities and research institutes in Victoria.

Tenders went out for the project in late November 2008 and a decision on a preferred partner will be made early this year.

In San Diego, Brumby said the idea of building a supercomputer had occurred to him many years ago, but a speech by Professor Peter Doherty triggered his renewed enthusiasm.

“When we were first elected to government and I was meeting with all of the groups about what we needed to do to elevate our position in the biotech space, we had a number of suggestions – a synchrotron, a stem cell centre – but a number of people said the next cab off the rank should be a supercomputer,” Brumby said.

“Peter Doherty made a speech about three or four years ago and he told me about all of his work at St Jude’s [Hospital in Tennessee] on children’s cancer and he said that if he had a supercomputer facility to analyse all of that data, they’d be able to further cancer diagnosis and cancer treatment. That had a big impact on me.

“So when the university formally made the application, I just thought: it’s too good an opportunity to pass up, and bang, we’ve done it.”

---PB--- eResearch

The University of Melbourne’s Professor Leon Sterling says the spark for the supercomputer proposal was ignited by a series of discussions with IBM about how to go about transforming science.

This is an area of great interest to Sterling, chair of software innovation and engineering at the university and also director of eResearch.

eResearch is a term that began to appear at the turn of the century as computer science began to realise that the only way to enable new generations of scientists was to create new enabling technologies, he says.

“[The term] was originally from the eScience movement in the UK,” Sterling says. “[Scientists would need] technology that would enable them to look at large-scale data using high-speed networks and high-performance computing. And to analyse the amount of data they were going to need social science.

“In the US they tended to use the term cyber-infrastructure and in Australian the term eResearch tended to be used. eResearch is inclusive of what it is that is going to enable researchers across the spectrum.

“Some people view it as a fundamental paradigm shift in the way that people do their research, while other people say it is just the necessary tool that augments their work.”

The main aim of the whole computational initiative, he says, is to transform the life sciences for researchers.

The new facilities will be complementary to others in Australia, such as ANU’s Supercomputer Facility, but the difference will be the quite remarkable concentration of life science researchers in Parkville.

Sterling admits that the University of Melbourne has been a little less proactive than others in providing centralised infrastructure, “but there has been a slow recognition that there is an argument for more centralised support”, he says.

“One of the things that characterises the Parkville precinct is the unusually large concentration of researchers, and I think that’s what the Victorian Government sees as well.

“Relative to other places in the world, to have that concentration of high quality international level research but less access to computational facilities … I don’t think the actual machine is going to be radically different but what will be different is the concentration on how to solve biological problems.”

The initiative involves a two-step process in which smaller machines will be purchased initially to build up to a faster machine in 2011-2012. The aim is to have the Victorian PCF rank amongst the top five in the world by then, and to be the top life sciences PCF in the world in 2013.

One of the major goals, besides the super-duper computer, is also to build a new generation of computational biologists in Australia, with the help of the preferred partner.

The partner will be expected to contribute its own staff expertise to building collaborative research here and overseas, and to locate some personnel at Parkville.

“Science is what is driving this,” Sterling says. “In order to do some of the transformative science, we won’t be able to do it without the computational resources of a high-performance environment.”