INTERCONNECTS

System Helps Researchers Study Role of Molecular Dynamics in Infertility, Thyroid Conditions: Researchers at the University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School have selected a high performance IBM computer to perform research they hope will one day lead to more effective treatment of medical conditions like infertility and thyroid problems. According to the Centers for Disease Control and Prevention, more than two million married couples in the U.S. are infertile; and many millions of Americans are affected by problems associated with the thyroid gland -- including weight gain, weight loss, depression, and fatigue -- though many may be unaware of the cause. The UMDNJ researchers are using the IBM system to create models of proteins and study the behavior of certain types of sensory molecules, called receptors, which regulate fertility and thyroid function. These receptors are related to a family of proteins that perform diverse functions ranging from detecting light and odors, to controlling blood pressure and regulating heart function. While the types of receptors being studied make up a large number of the molecules found in the human genome, and are common targets for drug treatments, very little is known about how they work at the molecular level. By learning more about this, the researchers at UMDNJ hope to discover how different types of drugs might be used to treat certain medical conditions more effectively. Using the high performance IBM system, which has a peak speed of one teraflops (one trillion floating point operations per second), the UMDNJ researchers are able to dramatically accelerate the pace of their work. Calculations that would have normally taken them a year to complete on a single workstation can now be done in just four days. Currently, the only way to view the makeup of a protein is through a 50-year-old process called X-ray crystallography, which involves crystallizing the protein and using a beam of X-rays to get an image -- called a crystal structure -- of the positions of the atoms within it. However, the receptors being studied at UMDNJ are membrane proteins and, due to the difficulty of crystallizing this type of protein, crystal structures exist for only two molecules of this entire family of related proteins. Furthermore, because crystal structures provide only a snapshot in time, it has been difficult for researchers to understand how these types of receptors work. To get around this problem, the UMDNJ researchers are using the IBM computer, an e1350 Linux cluster composed of System x3550 servers and 96 processors, to run a program called Amber that was designed by researchers at the University of California. Amber works by moving the simulated atoms of a receptor hundreds of thousands of times per second, many times over, until the researchers get a better idea of how the receptor is likely to react to other molecules. This process requires computational capabilities far exceeding those of traditional computers. "The type of research we're performing is very unique, and it requires some pretty mind-boggling calculations that could have never been done on our older systems," said Les Michelson, PhD, assistant director for special projects at UMDNJ. Also involved in the project is John Kerrigan, PhD, associate director of bioinformatics at The Cancer Institute of New Jersey, which is a Center of Excellence of UMDNJ-Robert Wood Johnson Medical School. He says, "With the power of our IBM system, we're gaining new insights into how these complex molecular switches known as receptors function. The computer models help guide experimental design in the lab as well as generate new hypotheses." "With its significant price/performance advantages and simplified management capabilities, the e1350 Linux cluster is the ideal high performance computing platform for the ambitious and important work being done by the researchers at UMDNJ," said Dave Turek, vice president of deep computing at IBM. The researchers have chosen to focus first on receptors that affect fertility and thyroid function because of the amount of biological information that already exists, and because of the prevalence of those problems. Additionally, because the receptors that control reproduction and thyroid function have features that are common to other receptors, the UMDNJ scientists hope their research will also provide insights into the functions of this large receptor family. This could help uncover better drug treatments for medical conditions unrelated to thyroid function or infertility.