Toronto’s Hospital for Sick Children (HSC) is now home to the most powerful supercomputer of its type in the world.
The 64-CPU Origin 2000 from Silicon Graphics Inc. is devoted exclusively to health and biological research in the public sector. It is scalable to 128 processors, features a CC Numa-based architecture,16GB of memory and close to a terabyte of disk space. About the size of three refrigerators placed side by side, it’s housed in its own climate-controlled room in the hospital.
Initially, the supercomputer will be used to search for similarities in gene sequences and to help develop three-dimensional models of human proteins.
There will be more human genetic data produced by scientists this year than in the entire history of mankind, said David Wharry, president of Silicon Graphics Canada Ltd. in Mississauga, Ont.. That is one reason systems like this are increasing in popularity.
“Gene sequencing for a human being consists of about three billion characters of data. If you were to print those out on pieces of paper and stack them, they would amount to about a third of the height of the CN Tower.”
According to Wharry, the Origin 2000 is able to process information about 1,000 times faster than a desktop system. Some of the more complex genetic problems take days to run on the Origin 2000, he noted. “So if you didn’t have this system, the same problem could take 40 or 50 years to solve.”
The computer, which cost about $4 million, was purchased from funds from an anonymous donor and a $1.4 million endowment from SGI. It arrived at the hospital in January and took about six weeks to install. Now, researchers from around the world access Genbank, a database of genetic data which resides on the supercomputer, said Dr. Jamie Cuticchia, head of the bioinformatics program at HSC.
“We are able to provide the fastest searching for DNA homologies in the world,” he said.
“If somebody has a piece of DNA they have got from an organism and they want to see if it is similar to anything that has ever been characterized in the past, we can search through gigabytes of information in literally a few seconds and return a result.”
The hospital is also using the supercomputer to design gene chips, which are small pieces of DNA placed onto glass or silicon matrices.
“[We] collect cells from different parts of the body, or cells that were exposed to different medicines, and then look and see whether or not certain genes are becoming overly expressed or inhibited.”
This information is then used to help develop drugs that will be safe for the genetic structure of humans.
“You are trying to kill a particular bacteria and you’ve done an experiment where you were able to interrupt genes inside this bacteria. And you’ve determined for 100 genes that if you disrupt them the bacteria dies. What you can do then is sequence those genes, then take that sequence and search Genbank to determine whether or not these (unknown) genes have ever been studied in other organisms,” he explained.
HSC previously used the main server at the National Center for Biotechnology Information (NCBI) in Bethesda, Md., to search Genbank. It took several hours to get results, Cuticchia said.
“Now, if we have 100 people working simultaneously – and I mean at that very second – all hit the return key and submit a search, it still returns the results in five seconds or less.”
Cuticchia admitted the idea of a such a huge system was intimidating at first.
“When we first bought a supercomputer, I had visions of HAL (the malevolent computer from 2001, A Space Odyssey) and these water-cooled things. But basically the architecture of supercomputers has simplified so much that if you know how to run a $10,000 Unix box, then you pretty much know how to run a $4 million Unix box.”
The supercomputer will change the way research is done because it will allow scientists to ask bigger questions, he said. “There’s a revolution going on inside of biology right now in the sense that it’s becoming more and more computationally complex. Before, a researcher could have a wonderful career working in his lab on one particular gene, asking one question after another. That doesn’t happen now – research labs look at hundreds of genes. Without the computational infrastructure to support that level of ‘big science’ you can’t be competitive.”
The computer has already attracted the attention of researchers from around the world, and HSC is in the process of recruiting a computational chemist, Cuticchia said.
“All of the [candidates] I have talked to during the recruitment process have said that one of the drawing factors of wanting to come to Sick Kids is the fact that we have this infrastructure in place already. They are not going to have to spend $40,000 start-up money for a computer that’s going to be one one-hundredth the speed of this one.”