Molecules in mid-air
A quite different project within the theme of human-computer intelligent interaction is the development of interfaces more effective than the keyboard and the mouse at manipulating complex data in a computer. Physics graduate student William Humphrey and research programmer Andrew Dalke are two members of the theoretical biophysics group, headed by physics professor Klaus Schulten, which is interested in developing simulations of the molecular dynamics (chemical or energy interactions) of biopolymers. The aim is simulations that are both accurate and intuitively understood and manipulated by the people studyng such molecules as proteins, DNA, or membranes
This dynamic modeling is extraordinarily complex. Each molecule may have anywhere from 3500 to 35 000 atoms, and the time scales of their internal processes range from a few femtoseconds to a nanosecond (or even up to a second in nature). Molecular dynamics simulations start with a computer model of a molecule in its initial condition. The equations of motion that describe the bonds, masses, and charges of atoms within a molecule are then applied, and the results of the simulation compared to the dynamics of its natural counterpart.
But as such simulations unfold, how can a mere human observer gain a quick and accurate idea of what is happening? To achieve that sensitivity, the Beckman team has developed interactive visualization tools to display the model molecules. A Silicon Graphics workstation with a projection system throws onto a screen two alternating views of a molecule, such as bacteriorhodopsin. Through special glasses, the molecule appears in its full three-dimensional glory as purple spirals curl around a central ball-and-stick model.
A scientist--or a visitor--can make the molecule translate from left to right or even come apart, simply by pointing with an index finger in the desired direction (two video cameras at right angles detect the gesture and transmit the information to the workstation). To view the molecule from a different angle, one calls out the command, "Rotate!" The molecule then slowly pirouettes, in reaction to a speech-recognition module. The whole human-computer interface blends Schulten's expertise in the molecules themselves as well as in visualization software, Thomas Huang's work on gesture recognition, Robert Skeel's development of software and algorithms, and Yunxin Zhao's background in speech recognition.
Helping one another see
Accessible by every Beckman Institute research project is the National Center for Supercomputing Applications (NCSA), part of which is housed in the Beckman Institute building. Through the center, institute researchers have access to such supercomputers as Thinking Machine's 512-node CM-5, Convex C3 and Exemplar, and SGI Power Challenge. Also located in the institute's building are the center's two Silicon Graphics workstation labs and virtual reality facilities, including the CAVE (Cave Automatic Virtual Environment). Named for the Allegory of the Cave found in Plato's Republic, in which the philosopher explored ideas of perception, reality, and illusion, this cubical virtual-reality theater measures about 3 meters on a side. Full-color stereo SGI graphics are projected onto three rear-projection screens for walls and a down-projection screen for the floor, with a resolution of nearly 3072 by 1536 pixels.
Computer-controlled audio surrounds a visitor with sound while the changing position of a user's head and hands are tracked with electromagnetic sensors clipped to goggles and fingers. The three-dimensional environment definitely eased the analysis of complex biological data, as became obvious when the task was to map the faint, thread-like tails of two sperm intertwined around a fertilized human ovum.
Central to the support of both the National Center for Supercomputing Applications and the three main research themes is the Visualization Facility on the Beckman Institute's fourth floor. The Viz Lab, as it is known, consists of the Digital Visualization Facility (which has many kinds of computers) and the Microscopy Suite (which has half a dozen state-of-the-art instruments). The Viz Lab is where the data gathered by researchers ends up for processing.
As designed by co-directors Bridget Carragher and Barbara Fossum, the Viz Lab is laid out in six "pods," each with two to four computer workstations, including Unix systems, high-end Macintoshes, and Pentium PCs. The workstations run a wide variety of software, ranging from large image-processing and analysis routines to animation packages. Carragher and Fossum designed the setup with the displays pointing inward toward the center of each pod, so that people are able to see each other's screens "to encourage multidisciplinary interaction," Fossum said. "People see someone else's good techniques and try them themselves and suggest them to others."
The scientists themselves sit and manipulate their data, perhaps calling for help from the Viz Lab's staff of consultants. "We're not just a service bureau," explained Fossum. "Every user works his own data, because that analysis in itself is research on the fly." Besides analyzing and displaying data for research projects, the Viz Lab gives researchers the horsepower to devise animated movies of data that change over time, either for research purposes or even for release to television news programs.
In addition to doing number-crunching, any of the workstations can be used to remotely observe and manipulate samples in two instruments in the Beckman Institute's basement: a transmission scanning electron microscope and a magnetic resonance imaging system. The instruments' control panels are accessed through Netscape or any other World Wide Web browser, so that they can be run from any place that has Web access.
In fact, the Viz Lab's co-directors are working with Paul Lauterbur, the inventor of magnetic resonance imaging, and others toward creating a World Wide Laboratory--the brain child of Clint Potter, the project's team leader. This lab eventually would also allow remote control of the scanning confocal microscope and atomic force microscope that also share the basement. If all goes well, a researcher from anywhere around the globe could "mail in a sample, a technician could put the sample into the requested microscope and walk out of the room, and the researcher could operate the instrument in real time via the Web," Fossum explained.
One stipulation made by Arnold Beckman in his institute's charter is common in industry but rare in academia. It is a quadrennial review: each research group must be scrutinized every four years to be sure it is contributing healthily to and benefiting from the multidisciplinary approach.
The official document, "Beckman Institute Research Review Policies and Procedures," minces no words about what this means: "The long term success of the Beckman Institute very much depends on whether the Institute can maintain its dynamic nature. Clearly, for a mature Institute this necessarily means that some percentage of faculty must return to their home departments in order to allow a healthy influx of new people."
"Beckman was very worried about [intellectual] petrification," Hess observed. Without some institutionalized way of stirring the pot, "the Beckman Institute would become just another building with people in it," Lyding added.
Then there's the practical issue of space. The institute cannot hire faculty directly. Every faculty member has to be tenured or on a tenure track in a home department. "But the existence of the Beckman Institute has been used as leverage [by the university] to hire the best," noted Jennifer M. Quirk, the institute's associate director for external affairs and research. Attracting anyone, however, is moot "if you're full and no one leaves."
Every program in each main research theme, therefore, must go under a microscope every four years to see if it is making the grade. This very month, on Nov. 18 and 19, the first-ever such review will be conducted. Up first will be the molecular and electronic nanostructures theme, chaired by Hess.
Of the review committee's six members, four will come from outside the university and two from the Beckman Institute's own Program Advisory Committee. The four external reviewers will be chosen not only for their research expertise and accomplishment, but also for "their demonstrated breadth of view and experience in interdisciplinary research."
Before the reviewers arrive on campus, they will read statements of the theme's mission, direction, and project work. Also included will be the rationale for why each group's presence in the Beckman Institute both receives and confers benefits. Then the six-member committee will visit the Beckman Institute for eight intense hours of presentations by faculty members, visits to laboratories, and meetings with the institute's director Jonas and associate director Quirk.
The review committee will then draft its evaluation of the theme and each of its composite programs. Their criteria will include its quality with respect to work elsewhere in the same fields, its effectiveness as part of the main research theme and the Beckman Institute at large, and its relevance to the larger goals of all three main research themes. On the basis of the committee's advice, director Jonas will issue a report in January 1997, listing his conclusions and recommendations as to which groups stay and which leave.
The molecular and electronic nanostructures is only the first main research theme to be examined. Next year, human-computer intelligent Interaction will come under the microscope, and in 1998 it will be the turn of biological intelligence. The year after that, it will be the National Center for Supercomputing Applications. Although the center is not part of the main research themes, "we want to have more of a relationship with [it] than as a tenant," Quirk added.
This pioneering review "has been talked about for a long time, but there is some resistance" to it, Quirk noted. Yet, all concerned agree that "if the Beckman Institute is to thrive, it needs to have new blood."
To Probe Further
For hot links to the Beckman Institute's World Wide Web home page and the page of the World Wide Laboratory and other research laboratories, consult IEEE Spectrum's own home page at http://www/spectrum.ieee.org.
A detailed description of the workings of a virtual reality CAVE can be found in "A 'room' with a 'view,' " by Thomas A. DeFanti, Daniel J. Sandin, and Carolina Cruz-Neira in the October 1993 issue of Spectrum, pp. 30-33.