Closing in on the DNA blueprint
Closing in on the DNA blueprint
CASE STUDY: DNA is the blueprint, but what makes cells grow and divide? At the Ludwig Institute for Cancer Research, answering this question involves protein research which could one day turn invasive cancer detection procedures into simply blood tests. To process their research results, they’ve taken on the powerful capacity of the recently launched Australian Proteomics Computational Facility; and they’re taking Intel, Dell and Microsoft along for the ride.
Often it’s not a lack of information but the lack of processing power and computing capacity that challenges the progression of medical research. At the Ludwig Institute for Cancer Research, a lack of processing power was preventing the efficient management of proteomic data, slowing down research on the proteins found in early cancer detection.
“For our cancer research, we were hamstrung by the lack of processing power that we were getting from smaller computers,” says Robert L. Moritz, director of the Australian Proteomics Computational Facility (APCF) at the Ludwig Institute for Cancer Research. “We had a number of servers around the place but they were all single processors that are also very expensive as well.”
After awarded a Federal Government grant for medical research, Moritz and his team at Ludwig worked to tackle these challenges through the development of a specialised centre to churn through protein related research. It took less than a year to get the APCF up and running, yet for the most part, its successful implementation rested firmly on the advancement of technology.
The mere collection of proteomic data is itself a major feat of endurance. The data is collected through ‘spectrometers’, designed for producing accurate, molecular weight measurements. But according to Moritz, processing data is pretty much all they do, while navigating the database is up to the user to negotiate.
“These machines produce thousands of these measurements per hour. To do that on a PC, a thousand measurements would take something like 5-6 hours to perform,” says Moritz. “When you’ve got several of these machines, you can immediately see the bottleneck that occurs.”
The problem with spectrometers is that their ability to collect data far outweighs the average workstation’s ability to process, analyse and house it. Moritz turned to processors and evaluated both what Intel and AMD had to offer. “We initially had a trial server using single core, then we evaluated the dual core processors from AMD,” says Moritz. “We were then offered the Quad-Core processors from Intel as a confidential study with them in mid 2006.”
The Quad-Core processors got straight to the heart of the APCF and produced what Moritz labels as ‘phenomenal results.’ Deciding to stick with the Intel processors permanently, the APCF wasted no time in having the processors delivered and installed earlier this year. “We ended up with 135 servers all up in a cluster, each of which were quad core, and 1080 processing cores in total,” says Moritz. “What we achieved, we never would have been able to achieve in a single or duo core situation, simply due to the limited amount of servers we could actually house.”
To mirror the results the APCF has already experienced in a duo core scenario, they would have needed to house double the servers - a cost Moritz says would have equalled the price of having twice the processing power. “With this, we also didn’t have to purchase the air conditioning, the additional power and of course the extra servers,” he says.
An even earlier technological partner on the project was Microsoft, who commenced work with the APCF back in 2005 and supplied the entire operating system for the cluster, the Windows 2003 cluster edition. On the hardware front, it was Dell who put all the bits and pieces together, by pioneering the engineering required on the system and finally putting the quad core processors into action.
In the field of proteomic research, it’s not just disease prevention the APCF is targeting, but also human and animal health. With up to 26 proteomic research groups currently utilising the APCF, the research examines anything from the early prevention of Cardiac disease, to producing more efficient crops for human consumption. “We’re building all these tools as generic tools for everyone to share no matter what proteomic tools they’re researching; there are at least 24 (research centres) across Australia,” says Moritz. “To build a facility just to do our own research wouldn’t have been feasible. We wouldn’t have been able to get the speed.”
Of course sharing the data across the country not only requires the processing power, but also the accessibility to make it happen. For that, the APCF deploys the use of a search engine known as ‘Mascots,’ which uses mass spectrometry data to specifically identify proteins from primary sequence databases.
At the Ludwig Cancer Institute the ability to speed up the identification of protein potentially involved in the progression of some diseases, could see the computing system saving lives in the years to come. “Colon, ovarian and similar cancers don’t have an early detection test, they’re all highly invasive and based on hospital stay and surgery which is difficult to apply to the population base on mass,” says Moritz. “Most of the people who become aware of the disease, become aware at a late stage. We’re trying to find the protein markers so we can start applying basic test and detect up to 90 percent of cases.”
By finally seeking out the capacity to process the large amount or raw data, Ludwig is working to improve on significant advancements in cancer research and make simple blood tests for cancer detection a reality. “DNA is a blueprint, but a blueprint doesn’t make a cell grow and divide,” says Moritz.
After being awarded a Federal Government grant for medical research, Moritz and his team at Ludwig worked to tackle these challenges through the development of a specialised centre to churn through protein related research. It took less than a year to get the APCF up and running, yet for the most part, its successful implementation rested firmly on the advancement of technology.
The mere collection of proteomic data is itself a major feat of endurance. The data is collected through ‘spectrometers’, designed for producing accurate, molecular weight measurements. But according to Moritz, processing data is pretty much all they do, while navigating the database is up to the user to negotiate.
“These machines produce thousands of these measurements per hour. To do that on a PC, a thousand measurements would take something like 5-6 hours to perform,” says Moritz. “When you’ve got several of these machines, you can immediately see the bottleneck that occurs.”
The problem with spectrometers is that their ability to collect data far outweighs the average workstation’s ability to process, analyse and house it. Moritz turned to processors and evaluated both what Intel and AMD had to offer. “We initially had a trial server using single core, then we evaluated the dual core processors from AMD,” says Moritz. “We were then offered the Quad-Core processors from Intel as a confidential study with them in mid 2006.”
The Quad-Core processors got straight to the heart of the APCF and produced what Moritz labels as ‘phenomenal results.’ Deciding to stick with the Intel processors permanently, the APCF wasted no time in having the processors delivered and installed earlier this year. “We ended up with 135 servers all up in a cluster, each of which were quad core, and 1080 processing cores in total,” says Moritz. “What we achieved, we never would have been able to achieve in a single or duo core situation, simply due to the limited amount of servers we could actually house.”
To mirror the results the APCF has already experienced in a duo core scenario, they would have needed to house double the servers - a cost Moritz says would have equalled the price of having twice the processing power. “With this, we also didn’t have to purchase the air conditioning, the additional power and of course the extra servers,” he says.
An even earlier technological partner on the project was Microsoft, who commenced work with the APCF back in 2005 and supplied the entire operating system for the cluster, the Windows 2003 cluster edition. On the hardware front, it was Dell who put all the bits and pieces together, by pioneering the engineering required on the system and finally putting the quad core processors into action.
In the field of proteomic research, it’s not just disease prevention the APCF is targeting, but also human and animal health. With up to 26 proteomic research groups currently utilising the APCF, the research examines anything from the early prevention of Cardiac disease, to producing more efficient crops for human consumption. “We’re building all these tools as generic tools for everyone to share no matter what proteomic tools they’re researching; there are at least 24 (research centres) across Australia,” says Moritz. “To build a facility just to do our own research wouldn’t have been feasible. We wouldn’t have been able to get the speed.”
Of course sharing the data across the country not only requires the processing power, but also the accessibility to make it happen. For that, the APCF deploys the use of a search engine known as ‘Mascots,’ which uses mass spectrometry data to specifically identify proteins from primary sequence databases.
At the Ludwig Cancer Institute the ability to speed up the identification of protein potentially involved in the progression of some diseases, could see the computing system saving lives in the years to come. “Colon, ovarian and similar cancers don’t have an early detection test, they’re all highly invasive and based on hospital stay and surgery which is difficult to apply to the population base on mass,” says Moritz. “Most of the people who become aware of the disease, become aware at a late stage. We’re trying to find the protein markers so we can start applying basic test and detect up to 90 percent of cases.”
By finally seeking out the capacity to process the large amount or raw data, Ludwig is working to improve on significant advancements in cancer research and make simple blood tests for cancer detection a reality. “DNA is a blueprint, but a blueprint doesn’t make a cell grow and divide,” says Moritz.