The power of CCNI can be used to complement any number of Rensselaer research projects. For example, in the signature research thrust area of energy and environment, von Maltzahn says faculty will use the supercomputer to create molecular-level models of the Hudson River or Lake George using data that are continuously collected from multiple mobile sensors immersed in the river or lake.
In other research areas, CCNI could be used to model social interactions between humans, or between humans and a digital persona or character. The center will also likely propel the research of Suvranu De, associate professor of mechanical, aerospace, and nuclear engineering, who is developing a “virtual patient” surgery training simulator. The program incorporates the sense of touch when using the instrument to poke a virtual patient’s kidney, for example, the surgeon would feel the appropriate touch feedback resistance and could lead to remote surgery technology that allows doctors to perform operations on patients who are in other parts of the world.
Two other projects already under way at CCNI will give the medical world a more thorough understanding of the human body and will make Rensselaer a more prominent player in bioengineering and bioinformatics research.
Kenneth Jansen, associate professor of mechanical, aerospace, and nuclear engineering with joint appointments in biomedical engineering and computer science, has been working to create a physiologically accurate model to track human blood flow. The model could eventually be used in hospitals to give surgeons insight into which of several procedures will be safest and most effective for any given patient.
When deciding what type of surgery to perform on a clogged artery, surgeons often rely on instinct and past experience when deciding to open up the artery with a stent, bypass the clog, or jump over the inclusion and connect the artery with another nearby. Both instinct and past experience are important in these situations, Jansen says, but everyone would benefit from that expertise being complemented by quantitative data.
Jansen, who is working alongside Shephard and other Rensselaer colleagues on the project, which is funded by the National Science Foundation, collects blood flow data and simulates the possible procedures to identify which one will best restore healthy circulation.
The model includes and keeps track of about 40 million different variables. Without supercomputers, Jansen says the project would be impossible. Five years ago, these simulations took a year to complete. Recently, the time was trimmed down to a day. The simulation scaled perfectly to the 2,048 processors of Rensselaer’s first Blue Gene computer, which cut the calculation time down to less than two hours.
Spreading the calculations across 16,192 processors should narrow the time to 15 minutes, Jansen says. He suspects that is a reasonable amount of time a surgeon would be willing to wait for the data to be analyzed, before performing an operation.
“I really see CCNI as an opportunity to demonstrate a concept,” Jansen says. “This was all interesting before, but the computers weren’t fast enough to do it. We can learn so much by having a tool that can produce such large volumes of information.”
Jansen’s colleague Angel Garcia, professor of physics, applied physics, and astronomy with a joint appointment in biology, who also heads Rensselaer’s Biocomputation and Bioinformatics Constellation, is also expecting to see an acceleration of his research into proteomics and protein folding, particularly protein aggregation. Understanding the process, which is a hallmark feature of neurodegenerative diseases including Alzheimer’s and Parkinson’s, could provide insight into methods of treating or counteracting it.
A few years ago, it took Garcia 200 days to run a simulation of the folding of a small protein. The protein had around 500 atoms to keep track of, but the model also has to recreate the 20,000 atoms or more of water in which the proteins are dissolved. Garcia condensed the computation time down to a week when he conducted research four years ago at Los Alamos National Laboratory, which at the time had one of the world’s top 10 supercomputers.
CCNI will cut that time even further. The desired result, he says, is a model that plays like a movie and can be stopped at any given split second to see where a particular atom is or how it is reacting to different scenarios. There are different methods of tracking the kinetics and thermodynamics related to protein folding, but all rely on supercomputers to handle dense computations to accurately represent key variables, including different solvents, pressure, and temperature conditions.
Garcia says CCNI’s computational power will enable him to investigate more complex interactions. The new wave of research should help propel Rensselaer to the forefront of academic biocomputation and bioinformatics research, and help to attract new faculty, researchers, and students.
Just as important as the research conducted at CCNI, says von Maltzahn, is the ability to take results from these virtual simulations and recreate them in the physical world. In some cases, such as protein interactions or synthesizing advanced materials, this can be done in Rensselaer labs. For more advanced fabrication, such as making semiconductors, the university must reach out to partners.
Kolb says the center’s second phase entails seeking out new research and partnership opportunities. Along with New York state and IBM, electronic design automation leader Cadence Design Systems and computer chip maker Advanced Micro Devices have already signed on to collaborate with Rensselaer through CCNI.
Each new relationship helps spread the word of Rensselaer’s groundbreaking research and facilities, and more partnerships in academia and with the private sector are imminent, Kolb says.
The astounding power of CCNI holds the promise of taking Rensselaer to a whole new level of research prominence, while helping to further realize The Rensselaer Plan goal of achieving global reach and global impact. “There are about 80 billion known galaxies in the universe,” Kolb says. “That means CCNI can perform about 1,000 calculations per second for every galaxy. To me that’s simply amazing.”