IBM COMPUTER GRANT WILL HELP RESEARCHERS LEARN MORE ABOUT THE BUILDING BLOCKS OF LIFE ON EARTH

RS/6000 Model 260 Workstations to speed computational chemistry efforts at the University of Arkansas

FAYETTEVILLE, Ark. -Two University of Arkansas professors have received a $250,000 grant from IBM for studies in two crucial research areas involving the computation of large molecules. The grant, in the form of four high-powered RS/6000 Model 260 workstations, is expected to dramatically reduce the time it takes to run complex computer simulations of molecular interactions - from one month to a single day.

IBM awarded the grant as part of its Shared University Research Program to Lothar Schafer, distinguished professor in the department of chemistry and biochemistry, and David Miller, associate professor in the department of crop, soil, and environmental sciences in support of their Biomolecular and Environmental Computational Chemistry Program.

Schafer will use the computers in his research on protein structures - the building blocks of life. Because proteins are the basis of life on this planet, understanding how they function has become increasingly important to scientists.

"To understand the chemistry of living cells, one must know the structure of proteins," Schafer said.

Researchers know that a protein's structure determines how it functions and interacts with other molecules. Scientists use X-ray crystallography to experimentally determine these structures but, since proteins are such complex and large molecules, the information obtained from the data is often not sufficient to obtain atomic level resolution. In this situation the results from computations help supplement the experimental information.

Schafer and his colleagues use computational chemistry to create a theoretical model of what a protein might look like. This can be done by first making an educated guess as to how the atoms in a given molecule might be placed. Then one can calculate the molecular energy as a function of the atomic coordinates for the resulting structure. The atoms are then shifted so the energy is lowered, to find the structure with the lowest energy, which is the most stable one.

While this process may sound straightforward, the calculations take lots of computer time and require large computers. When the calculations are performed at a quantum chemical level, the interactions of all the electrons in each atom are considered in the process. For a molecule with a thousand atoms, this requires executing billions of mathematical operations at each step of an energy optimization.

Schafer first began performing computer simulations of protein fragments in the early 1980s. In 1998, Schafer's group, working with researchers at the University of Antwerp in Belgium, published the first optimized structure of a protein ever calculated by a quantum chemical procedure.

"An exciting area of development in deep computing is the use of molecular-level information to investigate the properties of materials and complex organisms," said Bill Pulleyblank, Director of IBM's Deep Computing Institute. "Better understanding of these properties is transforming such diverse areas as automobile design and advanced pharmacology."

Computations of large molecules are also the basis of Schafer's collaborative work with Miller. Miller’s research seeks to determine how pesticides and other contaminants become immobilized in soils through absorption onto clay mineral surfaces. The two began working together when one of Miller's former graduate students, Brian Teppen, took a course from Schafer and became interested in the possibility of applying the techniques of computational chemistry to problems in soil science.

One of these problems is the retention of pesticides and other potential pollutants by soils. The molecular basis for this retention is the physical process referred to as adsorption. In soils, adsorption occurs most commonly on the surfaces of extremely fine-grained aluminosilicate minerals called clay minerals. The more strongly a molecule is adsorbed by clay surfaces, the lower its mobility, and the more difficulty it will have moving through the soil into underlying groundwater.

Miller and Schafer are attempting to simulate such adsorption reactions, with the ultimate goal of being able to calculate adsorption energies and thereby predict the potential of chemicals to move through the soil and contaminate groundwater.

"As scientists devise new herbicides, we hope to be able to model the new compounds and predict how strongly they will be adsorbed by the soil," Miller said.

The results of the computations can be compared to experimental data to make sure the computational procedures are accurate. In nature, the surfaces of clay minerals are believed to be coated with naturally occurring organic compounds collectively known as humic substances. The surface chemical properties of the clays, and therefore their tendency to adsorb pollutant molecules, is significantly altered by these humic coatings. Realistic computer simulations of contaminant adsorption reactions should therefore be performed using organically-coated, not pristine, clay surfaces. The exact chemical structure of the humic substances is not known, however, so the researchers are approximating humic-coated clays with protein-coated clays.

In their computer simulations, Miller and Schafer use a thick, plate-sized chunk of clay as a model for their studies. In that clay they typically study a few thousand atoms, some in proteins, others in pesticides and still others floating about as ions together with water in the soil. They take "snapshots" of the molecular interactions at intervals of a million billionth of a second. Such calculations take time, and, on current computing equipment, a single job requires more than a month to complete. With the new IBM RS/6000 Model 260 workstations, calculations that currently tie up Miller and Schafer's computers for 30 days will take about a single day to complete.

"We are very excited about this generous grant from IBM," said Mya Norman a graduate student in the department of chemistry and biochemistry. Together with fellow graduate student Ching-Hsing Yu, and research associate Susan Newton, she is in charge of the protein-clay project.

"The computational techniques have become so reliable that they can meaningfully interact with experiments and are really useful," Schafer said. "The University of Arkansas has always had a strong tradition in computational chemistry. Many of the techniques that are essential for performing successful molecular quantum chemical calculations were invented by my colleague, Peter Pulay, and are now used worldwide in all standard software packages. These theoretical tools have revolutionized the way work is done in chemistry."

# # #

About the IBM Shared University Research Program

The IBM Shared University Research ( SUR) program is designed to establish or enhance a strong IBM technical presence at selectedleading research universities and with leading researchers in the university community. In this program IBM grants equipment to universities in order to promote research in areas of mutual interest, and strives to connect the research and researchers at the university with personnel who are interested in the research from the IBM research, development and solutions provider communities. The SUR program at any given institution is well suited to initiate a strong ongoing relationship that benefits both the university as well as IBM.

About IBM RS/6000

More than 850,000 IBM RS/6000 systems have been shipped to over 125,000 commercial and technical customers around the world. The RS/6000 family of computers feature IBM RISC-based microprocessors and run AIX, IBM's UNIX operating system. RS/6000 delivers the industry's most complete UNIX offerings by combining applications with hardware, software, service and support - a combination that yields new levels of high availability, scalability, system management, performance, and deep computing capabilities. For more information, see: http://www.rs6000.ibm.com.

About IBM’s Deep Computing Institute

IBM's Deep Computing Institute was founded in May, 1999 with a mission to bring together experts in academia and industry to address some of the world’s most challenging business and scientific problems. Deep computing brings together very-large-scale computing power and highly efficient mathematical algorithms and software to develop solutions to very complex and difficult problems. More information on deep computing and the Deep Computing Institute can be found at: http://www.research.ibm.com/dci/.