The National Institutes of Health (NIH) contribution to the HPCC Program focuses on biomedical applications of computing and digital communications. Four NIH units have HPCC programs:
NIH participates in each of the five components of the overall Federal HPCC Program, as follows:
Through DCRT, NCI, and NCRR, NIH applies and evaluates scalable parallel computing systems to problems of biomedical significance. This includes adapting existing molecular analysis algorithms to new computing architectures, and developing entirely new approaches that take advantage of computational parallelism.
A cross-section of the three-dimensional reconstruction of a herpes simplex virus capsid is shown in the inset with the monoclonal antibody fragments bound to the protruding tips of hexons color- coded red and the original electron micrograph shown in the background.
DCRT and NLM have complementary investments in extending gigabit speed communications for high data volume scientific applications, as well as lower speed connections for a broad community of research and education institutions.
DCRT's efforts focus on high speed networking to support the intramural research program of the NIH; NLM serves as a national resource, providing support for medical centers to connect to the Internet, and developing prototype biomedical digital image libraries that use the Internet as a high speed distribution channel. An Intelligent Gateways project, co-sponsored by NLM and NSF, is developing methods to link dissimilar databanks over the Internet, using automated "Knowbots" (Knowledge Robots).
Each of the four NIH organizational units is developing algorithms and software for advanced, high performance computing environments. The two major themes of this work are molecular biology and biomedical imaging. Molecular biology computing includes comparison of genetic and protein sequences, and early development of algorithms to predict molecular structure and function. Biomedical imaging includes imaging of molecules and new methods for correlating and displaying clinical images in three dimensions.
In FY 1994 NIH will support the development of HPCC technologies for health care, with particular emphasis on the following applications:
Two to five research and development projects in each of these areas will be supported via an NIH Broad Agency Announcement contract mechanism issued in FY 1993.
NIH provides training and an ongoing investment in fundamental technology development. NCRR and NLM both sponsor formal degree-granting fellowship training in medical informatics, and cross-disciplinary training of established investigators in the use of advanced computing systems and methods. NCRR also supports a pilot project to educate high school science teachers and their students about new computing technologies for biomedical science.
The myoglobin protein with a hydration of 350 water molecules.
DCRT evaluated a 128 processor Intel Touchstone Gamma Prototype Parallel Computer and associated system software on biomedical problems in structural biology, computational chemistry, image processing, and genetic database searching.
DCRT designed and implemented remote file access and communication systems software for this computer.
NLM initiated a Medical Connections program in collaboration with NSF, to provide support for academic medical centers to connect to the Internet. Independently, NLM has provided support for a pilot project to connect rural hospitals in the Northwest to the Internet. This project is coordinated by the University of Washington Health Sciences Center Library, a Regional Medical Library in the National Network of Libraries of Medicine.
NLM began developing a system to provide Internet-mediated access to an electronic archive of 20,000 digital X-ray images for medical research purposes.
NLM created and published on CD-ROM a prototype Metathesaurus, Semantic Network, and Information Sources Map as part of the Unified Medical Language Systems research effort to link together computer-based biomedical information resources. These data files support experimentation on advanced systems that translate user information requests into multiple access vocabularies.
NLM initiated a demonstration project that will create a digital anatomy workstation that allows a health professions student to browse three-dimensional images generated by a specialized graphics computer at a distant site. The prototype links the NLM's Learning Center for Interactive Technology in Bethesda to the University of Washington at Seattle via the Internet.
NCI implemented remote file access to treatment guidelines from its PDQ (Physician Data Query) database via the Internet.
NCRR-supported High Performance Computing Resource Centers are evaluating high performance systems for biomedical research applications (see Side Bar). Investigators at these Centers have developed applications including:
Calculation of the binding of small molecules to DNA and other biological macromolecules depends upon a complex interaction of electrostatic fields, solvent-accessible surface, and other physicochemical properties. Structure-based drug design is a key biomedical HPCC Grand Challenge.
DCRT developed parallel methods for the following biomedical applications:
LOCATION VENDOR MACHINES Cornell Theory Center Kendall Square Research KSR-1 (64 nodes) IBM SP1 Pittsburgh Cray C-90 & T3D Supercomputer Center University of Illinois Thinking Machines CM200 at Champaign-Urbana University of Illinois Thinking Machines CM5 at Champaign-Urbana and Columbia University Columbia University Convex C210
NLM initiated a two-year project to acquire the three-dimensional digital representation of entire human beings at millimeter-level resolution, derived from computed tomography, magnetic resonance imaging, and digitized cryosections. This "Visible Human" research data set will become available nationally via the Internet in 1994.
Two-dimensional clinical images from computed tomography and magnetic resonance imaging underpin modern medical diagnosis. Transmission of these diagnostic images over wide area networks and their reconstruction to form three-dimensional views are important health care applications of HPCC technologies.
NLM created a prototype advanced molecular biology information retrieval program that provides integrated access to genetic and protein molecular sequences, and the biomedical literature linked to those sequences. Field testing of the system has begun.
NCRR and NLM achieved order of magnitude speedups in several existing molecular analysis algorithms.
DCRT and NCRR developed new algorithms for registration and rendering of three-dimensional images from two-dimensional clinical images and micrographs.
Research conducted at NCI's Biomedical Supercomputer Center is increasing the understanding of the human immunodeficiency virus (HIV) that causes AIDS and is helping to design and develop new drugs to combat the deadly disease. NCI researchers have successfully predicted the secondary structure of the entire 9,000 unit HIV virus RNA. NCI and NCRR supercomputing applications have assisted in the design of new drugs to inhibit HIV replication.
NLM competed the award of 10 Medical Informatics Training Grant programs at academic medical centers. The program supports cross- disciplinary training of health professionals in the use of advanced computing technologies.
NCRR conducted a pilot project to introduce scientific computing methods to high school science teachers and their students.
DCRT and NCRR sponsored "hands on" training of biomedical researchers in the use of new computational biology tools at NSF Supercomputer Centers and on the NIH campus in Bethesda.
In the first demonstration of its kind, scientists at a workstation in Chicago viewed high-resolution images of nerve cells in a high- voltage electron microscope located 1,700 miles away at the San Diego Microscopy and Imaging Resource (SDMIR), which is supported by NCRR. See Case Study 2 for further details.
Digital radiology techniques require high speed networks; a single X- ray film represented by a 2K-by-2K-by-10 bit gray scale generates a 4 megabyte image file.
NIH will accelerate the pace of molecular and genetic discovery by enabling the solution of currently intractable problems in molecular structure prediction, drug design, and human genome database analysis.
The program will apply and evaluate new computer architectures to key problems of human health and disease, in a manner that gives early feedback to computer designers on the strengths and limitations of their systems for medical applications. The HPCC Program will rapidly build an electronic community among life science researchers by connecting academic medical centers to the Internet. It will create prototype medical imaging applications that use the Internet and provide a model for distance-independent medical consultation. It will double the pool of computationally trained investigators in biomedicine.
DCRT will obtain a next generation parallel computer that will allow the solution of new computationally intensive problems in biomedicine.
NCRR will assess the efficiency and scalability of emerging massively parallel architectures for Grand Challenge problems.
NLM will establish Internet connections to 70 to 100 additional medical centers.
X-ray diffraction spectroscopy is a laboratory method for determining the folded structure of proteins and other biological macromolecules. Parallel computing systems can be used to automate interpretation of X-ray diffraction patterns acquired via two-dimensional array sensors.
NLM will create advanced three-dimensional imaging databases for digital anatomy on the Internet by the nation's health professions schools. Workstations to access and display those images will be developed.
An operational Knowbot-based database retrieval system will be deployed at NLM to provide integrated access to over 50 computerized knowledge sources in biomedicine, underpinned by a fully operational Unified Medical Language System that allows users to state scientific questions in their own language, and have the answer retrieved and synthesized automatically from multiple databases at multiple sites on the Internet.
NCI will expand its advanced software development program to allow research on a broader range of molecular dynamics, structure- function problems, and structure-assisted drug design.
NLM will develop and deploy advanced molecular biology workstations to approximately 1,000 molecular biology laboratories.
Through funding of five biomedical High Performance Computing Resource Center (HPCRC) programs at NSF and ARPA-sponsored high performance computing centers, NCRR will support development of algorithms to compare molecular sequences, predict molecular structure from genetic and protein sequences, simulate protein folding, and model complex biological systems such as proteins interacting with membranes in an aqueous solution. Computed images of biological structure, from molecular to whole body, will be an additional area of ASTA software development.
NIH will support the development of HPCC technologies through a Broad Agency Announcement to support two to five research and development projects in each of the six health-related areas listed above.
NCI will provide Xconf, a prototype multimedia (including medical images) group conferencing tool that uses the Internet. Contingent on funding, NCI will 1) connect to the Internet both medical research centers (to facilitate multicenter environmental epidemiology studies) and the Cancer Information Service offices at its cancer centers, and 2) begin developing telemammography over high speed networks.
In order to meet the needs of biomedical researchers, NCRR will establish at least one additional high performance computing resource center and upgrade the five existing centers. Enhanced cross training of biomedical research scientists will also be possible.
NLM will fund an additional 50 medical informatics training fellowships nationwide.