SCIENCE

Special-purpose supercomputer’s stay is extended, with new round of time allocations

Using a special-purpose supercomputer for biomolecular simulation, U.S. scientists have made significant advances in the understanding of protein function. The supercomputer, called Anton, designed by D. E. Shaw Research (DESRES), was made available to researchers through the National Resource for Biomedical Supercomputing (NRBSC) at the Pittsburgh Supercomputing Center (PSC).

Although DESRES has used Anton machines in internal research since late 2008, the NRBSC program marked the first time outside researchers used Anton.DESRES provided one of the machines without cost for non-commercial research by scientists at universities and other not-for-profit institutions. Motivated in large part by research advances in the first round of allocations, NRBSC and DESRES extended access to this resource beyond the scheduled end date of August 31, 2011.

In the first round of awards, announced in September 2010, a panel of experts convened by the National Research Council (NRC) of the National Academies of Science allocated time on Anton to 47 research groups. A $2.7 million “Grand Opportunities” grant to NRBSC from the National Institute of General Medical Sciences of the National Institutes of Health funded operation of the DESRES machine. The NRC panel will allocate a second-round of projects to both new and first-round awardees.

Access to the machine at PSC allowed researchers to execute ultra-fast “molecular dynamics” (MD) simulations that yielded new insights into the three-dimensional motions and biological function of various proteins over timescales far beyond the reach of even the fastest general-purpose supercomputers.

“Anton is an amazing machine,” said Martin Gruebele, of the University of Illinois, Urbana-Champaign.  Graduate student Yanxin Liu, collaborating with Gruebele and with UIUC biophysicist Klaus Schulten, has successfully simulated the “protein folding” process in a protein (lambda6-86) that, at 80 amino acids, is more than twice as large as the largest proteins whose folding had previously been successfully simulated and published.  “Anton enables simulations at full atomistic detail,” said Gruebele, “all protein atoms and water molecules included, for a long enough time for a protein to be folded from ‘first principles’ on the computer.  The simulations have revealed interesting dynamics in one of the alpha helices that wasn’t observed by experiments before, suggesting new experiments to probe into the folding mechanism of this protein.”

The ability of a protein with a given amino-acid sequence to fold into its characteristic three-dimensional structure is crucial for living cells; misfolded proteins not only lose their functions, but can also cause diseases, including Alzheimer’s and Huntington’s disease.

A goal of certain protein-folding simulations, Gruebele noted, is to be able, starting from only the amino-acid sequence, to arrive at the accurate 3D structure of the protein.  To be able to do this with computer calculations alone would represent a big leap forward in molecular biology.  The “magic number,” adds Gruebele, is about 200 amino acids.  “If we can do reliable simulations on that scale, we’ll be able by running on the computer to do what otherwise takes experiments and years of analysis.  Our work with Anton is exciting because it shows that this goal is within reach.”

In another Anton project, Emad Tajkhorshid, also at UIUC, simulated a family of proteins called membrane transporters, molecular passageways that open and close to move biomolecules, such as neurotransmitters, into and out of cells.  With Anton, Tajkhorshid simulated the structural changes in these transporters over a much longer period of time than has previously been possible.  “Before Anton,” he said, “we could simulate maybe 100 nanoseconds of protein motion. It’s a big advantage with Anton to be able to run several microseconds of simulation — more than 100 times longer in biological time.”

Tajkhorshid’s simulations show that a phenomenon, observed in some experiments, in which water can pass through a membrane transporter along with the “substrate” molecule is more widespread in this family of proteins than had previously been known.  “We are showing,” he said, “that this might be a universal phenomenon in the transporter superfamily.”  Their work points toward further experiments and suggests that the current understanding of membrane transporters may need to be revised.

Both Gruebele and Tajkhorshid are preparing papers to report their findings in scientific journals.  Other researchers using Anton have harvested unprecedented amounts of biomolecular simulation data that they are still analyzing.

Schulten of UIUC, in addition to his collaboration with Gruebele, simulated a complex process by which nascent proteins are threaded from the ribosome, where they are assembled, into the cellular membrane.  “Without Anton,” said Schulten, “our simulations were in the range of 100 nanoseconds; with Anton, we can now simulate between 10 and 100 microseconds.”  He has applied for additional time during the second round of allocations.

“Anton performs MD simulations up to 100 times faster than conventional supercomputers,” says Markus Dittrich of NRBSC, “making it possible for the first time to simulate the behavior of proteins over more than a millisecond of biological time.  The availability of these extended timescales has opened a new window on many important biological processes.”