David E. Shaw’s Supercomputer Is Uncovering Secrets of Human Biology

Anton is the supercomputer built by the legendary quantitative investment manager turned computational biochemist David E. Shaw and his research team. Shaw, who is also a senior research fellow at the Center for Computational Biology and Bioinformatics at Columbia University, spoke to a packed auditorium of Columbia Engineering students last spring about how the development of high-speed simulations is making possible insights into human biology that could lead to better understanding of protein behavior and to improvements in drug discovery.

Shaw’s affiliation with Columbia goes all the way back to the 1980s, when he taught as a young professor in the Computer Science Department after receiving his PhD from Stanford. In 1988, he founded the quantitative investment firm D. E. Shaw Group, which was among the first to apply algorithms to trading. Though known as a pioneer in the field of computational finance, since 2001, Shaw has shifted his primary focus to molecular biology through D. E. Shaw Research (DESRES), where he is chief scientist.

Shaw noted that computer science has been a common thread guiding his multifaceted career. A former teacher and mentor once counseled him to “stop trying to think of big ideas” and focus on the close study of a concrete application, since this often reveals fundamental problems whose solutions lead naturally to new, big ideas.

“What’s great about science and technology is that sometimes big, juicy theoretical discoveries grow out of practical work on specific applications,” Shaw said.

His initial foray into molecular simulations was inspired by the observation of a friend that important insights could be learned if simulations were fast enough to allow researchers to view molecule movement at longer timescales. At the time, such motions could only be viewed for less than 10 microseconds.

Shaw’s efforts at tackling this challenge culminated in Anton, a specialized supercomputer that can simulate the dynamic behavior of biomolecular systems orders of magnitude faster than other computational simulators. In fact, Anton is so fast that scientists can view molecular activity occurring over periods of up to 10 milliseconds. Such advantages allow researchers to study not just the structure of proteins, but their dynamics as well, along with interactions between proteins and drugs being delivered to the body.

Protein folding, in particular, is a rich area of study that has historically been difficult to explore because of researchers’ inability to monitor continuous structural changes in a protein over time in lab experiments. Anton’s ability to show protein folding is helping uncover what is involved in a number of neurodegenerative diseases such as Alzheimer’s and Parkinson’s, where the protein is actually “misfolding.”

Though Shaw believes the research is promising, he emphasized that simulation is not a replacement for experimental studies, and that bringing disparate methodologies and disciplines together often yields better results. His own team draws from computational chemists and biologists, computer scientists and applied mathematicians, and computer architects and engineers, while engaging in collaborations outside the team as well. DESRES has also made an Anton machine available without cost to outside academic researchers, and 150 research groups around the country have used the machine for their own projects.

Shaw, who is also an adjunct professor of biochemistry and molecular biophysics at Columbia University Medical Center, spoke to students as part of “Engineering Icons,” a series which brings to campus leading experts having a major impact in different fields of engineering and the applied sciences.