TORONTO — A young biotechnology company is running Linux on what may be Canada’s largest supercomputer to come up with faster ways to develop better drugs.
Medical Data Services Inc. (MDS) Thursday opened the doors on the research facility of MDS Proteomics, a subsidiary launched in January. The 52,000-square ft. building houses a series of IBM Unix servers in a 202-node cluster that Big Blue says is one of the top 10 non-military supercomputers in the world.
The MDS Proteomics supercomputer holds 15 terabytes of live data and can store 100 terabytes more in its archives so that scientists can compare the results of their research with what is already known about proteins. The company uses Linux, a technician said, so that it can easily customize the nodes in the cluster for specific tasks without worrying about operating system bugs. The company also has a backup and recovery centre with about half the capacity in Odense, Denmark.
MDS Proteomics will be using the equipment to speed up the process of analyzing the location and patterns of proteins in human cells. Scientists believe this field of study, proteomics, represents the next evolution in disease research following the mapping of the genome and could radically help pharmaceutical companies create more effective products.
The unveiling of MDS Proteomics follows close on the heels of a similar venture in Montreal, Caprion, which is using Sun Microsystems’ Sun Fire servers in its data centre. While corporate enterprises in many other vertical markets have cut back on major IT purchases this year, experts at the MDS facility indicated that life sciences represents an enormous opportunity for the deployment of high-performance computing equipment.
“With most drugs, we don’t understand how they work at the molecular level,” said MDS Proteomics president Frank Gleeson, “This is what sometimes leads to drugs being pulled off the markets and hopes being dashed.”
One of the tricky parts of developing drugs is finding the right proteins to “target.” There are about 500 known protein targets today. Companies like MDS and Caprion use supercomputers to vastly accelerate the number of proteins they analyze and map — sometimes researchers can accomplish in an hour what once took weeks — and to potentially find more targets.
“Suddenly we have all these choices,” said Dr. Eric Lander, director of the Whitehead Center for Genome Research and a professor of biology at MIT. “For a long time there was a lot of ‘me-too’ work in analyzing the same number of targets. Now we’re going to see the pharmas competing on the shrewdness of their choices.”
Indeed, creating a larger number of targets for drugs does not mean a flood of new products will hit the market anytime soon. Dr. Caroline Kovac, general manager of IBM Life Sciences, said that the process to get a drug to market can take 15 years.
Jessica Chutter, managing director and co-head of Biotechnology Corporate Finance at Morgan Stanley, said US$30 billion was spent on R&D in North America last year, but only 30 drugs were approved. This puts a severe strain on the biotechnology industry, she added, a market that was worth about US$112 billion at its peak and is now valued at US$50 billion today. This may be complicated by issues of intellectual property that emerge as technology facilitates more discoveries.
“When it takes that much money and time, you have to give the industry the capability to protect their work,” Kovac said. “The issues will largely be around what is patentable — how much knowledge has to be gained.”
Dr. Tony Pawson, one of MDS’s co-founders and co-director of the Samuel Lunenfeld Research Institute at Mount Sinai Hospital inToronto, said the short-term impact of proteomics research could be in the treatment of cancer and neurological diseases. Finding out where proteins are in the cell and why they’re there will significantly improve our diagnostic capability. “We’ve got a sort of handle on the genomic level . . . but that doesn’t tell us how the cancer cell is behaving,” he said. “Virtually everything we have is directed to a protein, but it’s been one at a time. Information technology is allowing us to ramp up considerably.”
In the future, Kovacs said it might be possible for biotechnology firms and pharmaceutical companies to design customized drugs or personalized medicine. “We get treated like statistics today,” she said. “Scientists look at symptoms, and when they see the same symptoms, that’s what get people classified as having a disease. That’s false, but that’s the best we can do today.”
