New Internet Capabilities Fueling Innovative Science

Editor's Note: This second part of a two-part series looks at the Internet's growing capabilities for scientists. For more and more researchers, the network is making crucial information resources available online. In addition, several ongoing demonstration projects in remote instrument control and in networked laboratories suggest a much-changed future for science as a result of the Internet. The first part of this series, which explored the increasing importance to scientists of E-mail and other electronic communication, appeared in the May 2 issue. The global Internet, with millions of individuals and thousands of computer networks, is changing fundamental aspects of the way scientists work. But the present, researchers agree, should be seen as prelude to a future in which an ever greater diversity of networked resources will allow scientists to approach new and currently unanswerable questions. Today, electronic mail and an array of other data exchanges move collaborative investigations forward that might otherwise be difficult or, perhaps, impossible, researchers say. Also, sophisticated online information resources are giving scientists ready access to needed journal and other databases. Tomorrow, researchers foresee, the Internet will offer still more to the growing number of disciplines dependent on the extensive network. It will support online, multimedia collaborations among scientists at different laboratories, for example, and sophisticated information retrieval from federated databases. The Internet of the future will also provide remote control of expensive or unique scientific instruments through what has been called "tele- experimentation." Distant instruments, in time zones around the world, will become increasingly accessible. "I can't think of any reason why people should go to [physically remote] telescopes, for example," says William Wulf, AT&T Professor of Engineering and Applied Science at the University of Virginia, Charlottesville. "Since most optical telescopes use [digital images] now rather than photographic film, there's just no reason to stay up late at night and look through eyepieces." Since serving as assistant director of the National Science Foundation in the late 1980s, Wulf has worked to promote the concept of the networked "collaboratory," a kind of common work space that makes the tools scientists use communally available through the Internet. "What's in this collaboratory?" Wulf asks. "I don't know, but there is a theme. Almost certainly, there is a common set of characteristics in terms of database support and collaboration technology to support interaction between people." As the Internet's capabilities grow during the 1990s, it will begin to provide simultaneous access to more and more databases of different types. As a result, scientists say, a kind of informational synergy will develop to put the answers to ever more complex queries within reach. This is especially true in the life sciences, some scientists assert. "The Internet is one of the absolutely crucial tools for modern biology," says Chris Fields, director of informatics at The Institute for Genomic Research (TIGR) in Gaithersburg, Md. "Biological research will become increasingly dominated by the exchange of large amounts of information and by cooperative work on large amounts of information. And the only way to support that is through networks." The degree of planning possible as this process goes forward is limited, researchers say, largely because of the decentralized nature of the Internet. While this fact may result in growing pains, they say, it is also one of the strengths of the network, ensuring flexibility and depth of expertise. "It is a grand sociological experiment," Wulf says. "And it's going to happen." Information And The Internet Databases accessible through the Internet are now central to many investigations in genetics, protein biochemistry, and related molecular biological sciences. Allied software tools to perform efficient searches of these databases are also available. Increasingly, these information resources are able to link different types of data together into broadly searchable federated databases. Sequencing data for a particular gene can be coupled with related literature references, for example, in a single search. The National Center for Biotechnology Information (NCBI) in the National Library of Medicine at the National Institutes of Health in Bethesda, Md., is the home of some of the most widely used of these biomedical informatics tools, including GenBank, MEDLINE, BLAST, and Entrez. In both the amount of information available and in the number of users, NCBI's information resources are growing at astounding rates, reflecting enormous biological research activity. GenBank, for example, is a database of all known nucleotide and protein sequences that includes supporting bibliographic and other information. According to James M. Ostell, chief of NCBI's information engineering branch, GenBank usage has increased tenfold since NCBI took over the database in October 1992 from an NIH contractor. In addition, more than half of the sequences currently in the database have been added since then. NCBI works to structure its information resources so that they are able to interact with each other as seamlessly as possible, Ostell says. This requires a great deal of initial interpretive work at NCBI as information from different sources is integrated into the system, but the result is a kind of unified "information space." "From the user's point of view, this is a very large integrated space with many different kinds of information all linked together and available all at once," Ostell says. "In reality, it's many different databases running in the background. The service knows how to get from one to another in a smooth way." Ostell adds: "There are a couple of buzzwords floating around. Collaboratory is one, and there's another one called the federated database approach. They're all groping after the same thing, this idea that you can have different groups of people working in overlapping projects and have some way, through computers, to get a view that includes all their different projects together." Although NCBI is responsible for putting several important information resources for biologists on the Internet, there are other groups--and individuals--contributing, too. One key to the Internet's growth and success with scientific users overall has been its distributed, nonhierarchical nature. Put another way, the tools and information available to the biological research community are also produced and maintained by that same group. Researchers at TIGR, headed by J. Craig Venter, are developing a pair of powerful databases that they expect to make network-accessible by early summer. TIGR is an independent, nonprofit research laboratory using Venter's expressed-sequence-tag method--based on automated partial sequencing of cDNA--to hunt for new genes in humans and other organisms. According to Chris Fields, TIGR's informatics chief, the databases will reflect ongoing investigations at the lab. The Expressed Gene Anatomy Database (EGAD) will include gene expression and function data; and the Sequences, Sources, and Taxa (SST) database will incorporate phylogenetic, specimen, and collection data. TIGR's databases will contribute to the Internet's capacity to help researchers find answers to increasingly complicated questions, Fields says. The Internet is the only reasonable way to approach such a goal, he says, with distributed, federated information resources being "curated" by the research community at large. With a museum collection of specimens, Fields says, the curator must constantly add new information about the specimens and relate them to each other and to specimens in other collections. "The same is true of data," Fields says. "But you can't expect one place to have anything like the expertise required to do the curation. These resources will be produced by whole communities with entirely different sets of expertise. And they need to be curated continuously by those communities." Fields adds: "With good federated database technology, all of that can happen." The Coming `Collaboratory' Many scientific instruments are now controlled by computers, and their output is digitally acquired and processed data. Usually, the computers are an integral part of the instrument physically or, at least, in the same laboratory facility. In the process of developing a million-dollar, one-of-a-kind 400 kV transmission electron microscope for neuroscience applications, however, researchers at the Microscopy and Imaging Resource of the University of California, San Diego, realized that there was no necessity for this. Computer control of the unique microscope also meant that it could become an early entrant into what they call the "distributed laboratory." It could be made available, over the Internet, to neuroscientists here. A crucial second element in putting the microscope online, according to Mark Ellisman, a professor of neuroscience at UC-San Diego who heads the project, is being able to process the imaging data within the time span of a reasonable work session. The three-dimensional data sets generated by the microscope are huge and require considerable image processing and computation before viewing--seven to 10 hours on a powerful workstation. For this reason, his group linked the microscope to the San Diego Supercomputer Center across campus, where images can be returned in about 20 minutes. "There are a lot of researchers who would like to collaborate with us," Ellisman says, "but might not always be able to fly out here. We make it possible for them to send a specimen, for us to map it, and then, with software we provide that allows them to look at the mapped specimen, for them to drive the micro-scope remotely." For one study involving an animal model of the neural degeneration associated with Parkinson's disease, collaborating investigators are located in Memphis, Tenn., and Edinburgh, Scotland. Interest in the microscope, currently a demonstration project funded by NSF and NIH, has been growing, according to Ellisman. "I wasn't sure that the community would be that interested in [the project]," Ellisman says. "I am now." At the Pacific National Laboratory (PNL) in Richland, Wash., nuclear physicist Richard Kouzes is working to establish a collaboratory on the model put forth by William Wulf. The capacity for "tele-experimentation"--remote control of experiments through the Internet--is one goal of the collaboratory, Kouzes says. He organized a workshop held at PNL in March to explore how the Environmental and Molecular Sciences Laboratory, a $230 million effort scheduled for completion in 1997, might best incorporate sophisticated networking. Kouzes hopes, through the Internet, to be able to link researchers in an interactive group by better integrating a number of existing software tools, some specifically de- signed to enable online collaboration. The graphical user interface for the Internet called Mosaic, developed and distributed free of charge by the University of Illinois, Urbana-Champaign, is one such collaborative tool, he says. Another is called MBONE, available--again, at no charge--from the Lawrence Berkeley Laboratory, Berkeley, Calif. MBONE, which stands for multicast backbone, allows sophisticated audiovisual teleconferencing among researchers at different sites over the Internet. One feature of MBONE is a "whiteboard" area on the screen where a researcher can display pictures, graphics, or text for simultaneous viewing by all conference participants. "It's a common work space, if you like," Kouzes says. Another software tool that is likely to be useful in the collaboratory, Kouzes says, is the multimedia Virtual Notebook System (VNS), available from the ForeFront Group Inc. of Houston. VNS allows geographically separated researchers to contribute information to a shared "notebook" through the Internet. Researchers at Cornell University, Ithaca, N.Y., have developed a similar tool called The Scientist's Workbench, Kouzes says. "The whole concept of the collaboratory is that it doesn't have a geographical location," Kouzes says. "It is, rather, everywhere."

New Internet Capabilities Fueling Innovative Science

Interested in reading more?

Become a Member of