Computers Seen Revolutionizing Research Labs

Three computers - an IBM PC and two Sun workstations-sit on the desk of molecular biologist Richard Roberts, assistant director for research at Cold Spring Harbor Laboratory on Long Island. Roberts routinely uses all three in the course of a day, moving easily from word processing to a search of DNA and protein databases to a display of sequences, and to running simulation models of DNA methylases, special proteins that stop restriction enzymes from cutting bits of genetic material. "I look at computers from a practical standpoint," he says. "I want to use them as tools." But for Roberts, three computers are not enough. He longs for access to the connecting software and networks that would make it possible for him to work in a fully functioning electronic environment. If Roberts wants to check the scientific literature for previous work on a particular chemical structure, for example, he must walk to the library. Although some institutions allow scientists and students limited computer access to their library systems, that's not the case for Roberts. And even when it's possible to conduct off-site searches, often researchers are limited to locating a book or journal rather than obtaining its contents, especially if the information appears in an obscure publication. If the library doesn't carry the journal they want, it could take weeks for them to obtain it through an interlibrary loan. What: The Seventh Annual Scientific Computing and Automation Conference and Exposition When: September 24-26 Where: Philadelphia Civic Center Who: About 4,000 scientists expected, primarily from the United States Keynote address: "Investigating Scientific Fraud," a review of the facts in the David Baltimore case, by Walter W. Stewart, National Institutes of Health, Tuesday, Sept. 24, 9 A.M. Among the more than 50 sessions on basic research, current applications, and future trends in scientific computing will be: * "Visual Computing in the 1990s," Tuesday, Sept. 24. * "Information Erector Set -- Building Your Own Scientific Information System," Wednesday, Sept. 25. * "State-of-the-Art in Electronic Notebooks," Thursday, Sept. 26. "These are all things that could be done quickly by a sensible computer program," Roberts says about his unfilled data needs. "You can do the science in 10 seconds, but to take two weeks to search the literature to find out what it means is kind of crazy." Roberts's predicament is evidence that problems remain to be worked out before the computer becomes an infallible lab partner. But researchers say that in recent years the expanding technology has improved their ability to do more and better science. The next generation of computing and communications technologies being developed promises to rescue Roberts and thousands of other scientists from this aggravating situation. Computer experts say that easily accessible, extensive digital libraries will be only one of a panoply of information tools available to scientists. Other essential components in the scientific computing environment are databases, an assortment of computing options for number crunching and data analysis, better networks for more efficient communication with colleagues over long distances, and tools to make possible real-time visualization of data. "There will be a great democratization of access to the computational means of production," says Larry Smarr, an astrophysicist and director of the National Center for Supercomputing Applications at the University of Illinois at Urbana. "It will empower individual-investigator small science wherever that investigator is. At the same time, it will create a new type of collaborative, interdisciplinary science that simply doesn't exist today." Some of these tools are already on the market. Next week, at the seventh annual Scientific Computer and Automation Conference and Exposition, researchers will be able to learn more about such key elements as scientific visualization, electronic laboratory notebooks, high-speed networks, and the latest techniques for building research databases. The show runs from September 24-26 in Philadelphia (see box). Investigators who are leading this revolution in the scientific computer environment often refer to their work setting as a "collaboratory." The term was coined by William Wulf, former head of the computer and information science and engineering directorate at the National Science Foundation and now a professor at the University of Virginia, and by Laurence Rosenberg, who is leading a multidisciplinary program at NSF to design better computer and communications systems. "A collaboratory is a place in which people interact with instrumentation, with data, with journals and books, and with each other," wrote Wulf and Rosenberg in an article for EDU Magazine, published by Digital Equipment Corp. (DEC) of Maynard, Mass. (Spring 1990, page 2). "Except that it's not really a place--it's a distributed electronic environment." According to computer experts, several new technologies will be the bedrock of the collaboratory. Here are brief descriptions of some of them: "A gleam in the eyes of many scientists for a number of years," according to DEC researcher Richard Lysakowski, electronic notebooks are simply a computerized version of the pen and paper now used to keep a log of methods, results, analysis, and thoughts that scientists encounter on their path toward discovery. Like a paper notebook, an electronic version could produce a legal record of what transpired in the lab. It most likely would consist of a laptop computer, much smaller than today's versions. Information entered into the notebook could be accessible by date, subject, and other relevant categories. Another feature of the electronic notebook would be its integration into the works of other scientists. Researchers from anywhere in the United States could connect their notebooks and compare data simultaneously as they ran their experiments. Some forms of electronic lab notebooks are already on the market, including word processors with templates for data entry. Recent technological advances will yield more sophisticated models, but legal and other issues remain to be solved. "A lot of it is vision right now," says John Bennett, manager of scientific marketing at DEC. "The computing technologies and the environment have just become sophisticated enough to let people begin to think about the possibilities." Scientists need prompt and efficient access to a wide range of constantly expanding databases. Some of these databases are accessible today on Internet, the national research network. But such files as the genomes of various organisms and the names and coordinates of astronomical objects, several of which are already available, are only the beginning of what will be needed in the decades ahead. "As we get into more and more large-scale projects, we're going to need transparent interconnections that keep track of what's being found, who did it, and how it was done," says Leroy Hood, director of the Center for Molecular Biotechnology at the California Institute of Technology. Developing these and other computational tools, he says, "will be critical for the further development of biology over the next 20 years." To develop distributed databases on global change, for example, researchers at a number of campuses of the University of California joined this summer with DEC to form Project Sequoia 2000. The $15 million project will develop computer hardware and software to manage massive amounts of data generated in the study of phenomena relevant to the environment, including global warming, ozone depletion, deforestation, and contamination of the world's water supply. Some universities have developed online catalogs, abstracts, and other material that can be accessed from off campus, and organizations and private firms are beginning to introduce electronic journals (The Scientist, March 4, 1991, page 4). But the economics of the enterprise may pose as much of a problem as does the technology, say some scientists. "There's been a failure of libraries and software companies to figure out how they can be financially comfortable [with electronic publications]," says Roberts. At the same time, Roberts says he hopes that such problems do not deter more companies from putting more of their material online. "The ability to search the literature by computer would just make an enormous difference," he says. At many large research institutions, the scientific computing environment already links a complex assortment of machines, from Macintoshes to the most powerful supercomputers. For example, investigators at Caltech have collaborated to create special "supercomputational chips" that can do rapid searches of patterns within specific DNA sequences. "These let small computers do calculations that supercomputers normally would do," says Hood. Although such tools could one day make high-performance supercomputers obsolete, say some scientists, a more likely result will be to give researchers more options. "You'll be able to spread your problem across the environment and use it to address the problem in ways that make sense," says Norman Morse, head of the computing and communications division at Los Alamos National Laboratory. "You will choose the most numerically intensive parts of your problem and run those on the high-performance end," leaving other work for a less powerful machine. As data from computations and other forms of investigations proliferate, researchers are beginning to use visualization techniques to take advantage of the efficient visual computer in our brains. Rather than trying to make sense of millions of numbers generated in a blizzard of computer printouts, scientists are using specialized workstations to change the numbers into meaningful pictures that can then be assembled on a videotape and analyzed. For instance, astrophysicists studying the pattern of pressure waves created when a stream of gas smashes through the boundary between two galaxies can start a computer simulation with given values of temperature, pressure, and other variables over a two-dimensional grid. A supercomputer solves equations that describe the properties of the gas "jet" and the media it travels through. The result is a massive set of data points at each step in the simulation. The researcher assigns different colors to the spectrum of pressure values--with, say, red representing the highest pressure and blue the lowest. Specialized graphics machines and software construct an image of pressure distribution throughout the grid. "Visualization is the hottest area in terms of broad applications," says DEC's John Bennett. "The field has matured a lot in the past few years. It's becoming more interactive." Connecting the advanced elements of the scientific computing environment are networks. They are becoming faster, easier to use, and more available. Among the most talked about is the National Research and Education Network (NREN), which within the decade is expected to connect millions of researchers at more than 1,000 institutions. Its backbone will be a band width of several billion gigabits, sufficient to transmit about 100,000 pages of text each second. High-speed networks are being tested in various parts of the U.S., conveying a broad range of raw and processed data to researchers. "As we in the computer field know all too well, such large quantitative changes often result in significant qualitative changes as well," wrote Wulf and Rosenberg. "But it's difficult to predict what those qualitative changes will be." With the use of electronic mail skyrocketing among researchers in academia, government, and industry, various tools are being introduced to help scientists within these distributed computing environments work together. These include what has become known as groupware, software that helps scientists in different locations work on a problem simultaneously. "Many of us believe that this area of collaboration research represents the most significant potential breakthrough technology for the real world," says Bennett. "For the last 30 years, computers have done a lot to make calculations more efficient. What they haven't done, however, is to help people work together better. This must include people who are not computer literate." Illinois' Smarr can envision several scientists, sitting at workstations around the country, monitoring graphic output from an ongoing supercomputer calculation. In Smarr's vision, as the data are stored in files on a remote mainframe computer, the scientists watch it on one window of their workstation screens and discuss its implications via a video teleconferencing window. At their fingertips is the entire sequence of calculations relating to the problem. As the discussion progresses, scientists can call up pertinent graphics, enlist the help of colleagues, or access a key paper from a remote digital library--all without leaving their chairs. "The point is," Smarr says, "to create a comprehensive digital interactive visual environment in which all resources required to do science are present simultaneously." William H. Allen is a science writer for the St. Louis Post-Dispatch.

Computers Seen Revolutionizing Research Labs

Interested in reading more?

Become a Member of