David Shaw had established one of the leading quantitative hedge fund firms by the late ’90s, with a decadelong record of investing excellence, but still the former computer science professor felt unfulfilled. “From a scientific point of view, I felt like I was getting stupider with every passing year,” he says. “I was forgetting things I’d learned in undergraduate calculus. I found it really depressing.”
Shaw’s epiphany came in 2001, at his 50th birthday party, when he spoke with the late William Golden, then 91 and a leading figure in U.S. science policy since the 1950s. Golden, whom Shaw had met shortly after being appointed to the President’s Council of Advisors on Science and Technology in 1994 by Bill Clinton, told the hedge fund manager to follow his heart.
Has he ever. Over the past seven years, Shaw has assembled a team of 75 biologists, chemists, computer scientists, engineers, mathematicians and physicists and built D.E. Shaw Research into a leading innovator in the field of molecular dynamics, which uses computers to create 3-D images of biological structures.
The inspiration for the research arm harks back to Shaw’s days at Columbia University in the mid-1980s, when he was developing a massively parallel supercomputer for use in artificial-intelligence systems. Cyrus Levinthal, then chairman of the university’s biological sciences department and a pioneer in molecular dynamics, had designed a software program that tried to predict how proteins assemble themselves, a process called folding, but he needed a specialized supercomputer to run it. Shaw couldn’t come up with a solution at the time, but he found the problem interesting — and, ultimately, irresistible. When he decided to step back from the day-to-day grind of the hedge fund business and return to the world of scientific research, he chose to tackle the protein-folding problem.
“At a certain point I started thinking it might be possible to build a supercomputer that could shed light on the folding process and on other biologically important phenomena by simulating the dynamical behavior of proteins,” Shaw says. “But you’d have to design such a machine in a completely different way than a regular computer.”
That’s what he and his research team have done. Last fall they unveiled Anton, a massively parallel supercomputer that can perform very long, accurate molecular simulations — crunching enormous amounts of data to create virtual movies of molecules at an atomic level of detail. Anton and the research venture are housed on the 32nd and 33rd floors of Tower 45, the Manhattan headquarters of the hedge fund firm D.E. Shaw & Co. Shaw named the computer after Anton von Leeuwenhoek, the 17th-century Dutch scientist and inventor known as the father of microscopy, and has spent in the tens of millions of dollars on the project.
“Anton is probably David’s boldest move,” says Klaus Schulten, head of the Theoretical and Computational Biophysics Group at the University of Illinois at Urbana-Champaign, which has developed its own supercomputer for doing molecular simulations.
To represent a molecule in motion, Anton must calculate the forces acting on each of its atoms, incorporating potentially hundreds of years of data from chemistry experiments and the laws of physics. Shaw and his group designed Anton’s 512 specialized processors, which were built by Japan’s Fujitsu, as well as the sophisticated software that together generate the simulations. Shaw plans to build 16 more supercomputers, and hopes to make one available for free to universities and other not-for-profit research labs.
The work by D.E. Shaw Research (and by Schulten’s group at the University of Illinois) has the potential to revolutionize drug development. So-called computational microscopes like Anton could enable researchers to design drugs at the molecular level rather than through trial and error. Shaw points to work his group has done on the protein c-Abl kinase, which when it goes awry causes chronic myelogenous leukemia.
“We’ve tried to learn more about how the protein moves and changes shape, what causes it to do that and what functional significance those motions might have,” Shaw explains. “If a mutation causes it to get stuck in its active state, the protein can cause a life-threatening form of cancer. Chronic myelogenous leukemia has recently become highly treatable using a drug called Gleevec, which has significantly extended the lives of many patients, but most of them eventually become resistant to the drug’s effects. If you understood the biology really well and could visualize in detail how this drug is binding to the protein and changing its behavior, the hope would be that somebody might ultimately use that knowledge to design new drugs that attack the disease in other ways.”
As the chief architect of Anton, Shaw sees his role in part as that of an enabler for other, even greater breakthroughs: “I have no illusions of being a general in the war against cancer, but it feels good to be serving as a soldier.”
