Tools for Collaboration
JAMES D. MYERS
PACIFIC NORTHWEST NATIONAL LABORATORY
At Pacific Northwest National Laboratory we have created a collaboratory at the Environmental Molecular Sciences Laboratory (EMSL), a facility that opened its doors in 1997. EMSL encompasses about 200 researchers, some 70 laboratories, and one of the world's fastest supercomputers, all basically doing molecular-level research to enhance our understanding of environmental issues. The EMSL is a U.S. Department of Energy (DOE) user facility. About half of the time on EMSL instruments is available to outside users. EMSL is located in the middle of a desert in eastern Washington and is fairly difficult to reach. Given this challenge, providing an electronic means for outside users to do research there, and for onsite researchers to collaborate with outside scientists, is a priority. For these reasons and to support EMSL's interdisciplinary approach, we have wholeheartedly adopted the concept of electronic collaborative research.
Much of the research that EMSL does that is related to collaboratories is done in partnership with many national laboratories, universities, and companies, as a participant in DOE's DOE 2000 Project. EMSL is also collaborating with the National Center for Supercomputing Applications at the University of Illinois, Urbana-Champaign to develop real-time electronic tools for collaborative research and the Collaborative Electronic Notebook Systems Association, a consortium of chemical and pharmaceutical companies to promote electronic notebooks.
One direction for the development of collaborative tools is to support real-time interactions. Such tools include video conferencing, shared browsers, whiteboards, chat boxes, and shared applications. These technologies enable participants to discuss data, draw sketches directly on a whiteboard that others can view, call up data in their visualization applications and rotate and zoom-in on images, and easily guide each other across the Web. Remote camera controllers allow users to pan, tilt, and zoom the camera around a room, focusing on a lecturer or members of an audience if desired. In the laboratory it can focus on equipment or view experimental procedures.
Another new technology is an electronic notebook, the equivalent of a bound paper laboratory notebook that enables scientists to use the Web to post text, images, data files, and graphs for colleagues. Anyone with a modern computer, a good Internet connection, and an inexpensive camera and echo canceller can
collaborate using these tools. We may one day collaborate in three-dimensional immersive environments wearing stereo goggles, but no such fancy equipment is needed to collaborate remotely today. As scientists use Internet collaboration tools more frequently, they will become integral to the overall research environment.
Before deciding which tools to use in their work, researchers first need to consider what occurs when they do science and how collaboration can help. Setting up a collaboratory is not simply a matter of running a remote experiment. Remote control software may let participants perform the experiment, but they will also need access to the sample preparation procedures, instrument settings, and other information usually recorded in a local paper notebook today. Before the experiment can be considered, potential participants must discover the remote resource, understand its capabilities, contact the local researchers, develop trust, and perhaps receive training on a remote instrument. Even if the researchers decide to visit the EMSL to conduct the actual experiment, they can meet people, understand procedures, and learn about the instrument before they arrive. Remote researchers must also find effective techniques for analyzing the data and consulting with coresearchers in writing up publications. Because scientific data are often complex and multidimensional, researchers will need to be able to confer with local researchers familiar with analysis of data from EMSL instruments.
Taking all such tasks into consideration allows one to identify the suite of tools that can best help facilitate the collaboration. Because every scientist does unique work and experiments are by definition at the cutting edge, scientific collaborations do not lend themselves to a cut-and-dried business or assembly-line approach. No one can say, "I want the latest quarterly numbers so I can run them through the same spreadsheet I ran them through last week" and work only by mailing data files around. Instead, researchers must be able to confer, analyze preliminary data, develop hypotheses, perform additional experiments, and think together in a very dynamic, exploratory fashion.
Complicating the process even further is the fact that the collaborative part of research can be intermittent. A scientist may take six months to build a new piece of equipment and only then be ready to collaborate. Collaborative tools need to be simple so that once people are trained to use them, they do not forget everything by the time they are ready to collaborate.
When we started developing the EMSL collaboratory about five years ago, we first interviewed researchers in the laboratories to understand their needs. This information enabled us to divide collaborations into four general types that involve different modes of communication. The first is peer to peer, in which two people work in the same discipline, often wanting joint remote access to an instrument to obtain raw data. The second, mentor-student collaboration, requires much more interaction. One person often demonstrates a technique as the other watches and then observes as the other tries to replicate it. Interdisciplinary research, a third type, entails many of the same interactions as mentor-student collaborations. Each participant brings a certain base of knowledge to the experiment. As the partici-
pants work together, each becomes the student of the other in their respective fields. The final form of collaboration involves what I call producer-consumer work, in which people performing the experiment are looking for results and nothing more. A biologist may not want to know how a specific research technique such as mass spectroscopy works, nor do they want to operate such an instrument remotely. They may simply want their collaborator to determine a protein's sequence for them. Thinking about collaboration in these terms helps us understand what tools—remote instrument control, conferencing, whiteboards, electronic notebooks—will be most helpful for a particular group.
To create a collaboratory involving EMSL's nuclear magnetic resonance (NMR) facility, we initially provided basic tools, including video conferencing, whiteboards, and an electronic notebook to one group studying the three-dimensional structure of a particular protein. We also gave the remote researcher access to the NMR spectrometer control software via the Internet, allowing him to run experiments remotely. While we did not have to change the spectrometer software to do this (it uses the X-Windows protocol), we did have to deploy additional software that encrypted all communications between NMR spectrometer and remote user to ensure that unauthorized users could not gain control of the spectrometer.
We watched how this group collaborated over the Internet and then started looking at what else we could do to help them. We saw that the researchers were using a very tedious process to share the experiment parameters (nearly 20 pages of voltages, frequencies, time delays, amplifier settings, etc., that a local researcher would usually print a copy of). They would save the parameters as a text file, open the Web browser, log-in to the notebook, upload the text file, and then view it as one long scrollable page. We made this much easier by writing software that allowed the parameters to be sent to the notebook directly from the instrument control software at the click of a button, with no need to start a browser or go through the notebook log-in screen. We also designed an extension to the notebook that allows users to easily search for a given parameter. Rather than scrolling through pages of text, users can now type in the first couple of letters of a parameter name, or choose one from a list, and see its value in a simple text box. By making these changes it became easier for both the local and the remote researchers to work on the electronic notebook than on paper.
Another enhancement we made involved integrating Java software developed at the European Molecular Biology Laboratory (EMBL) that can display Brookhaven Protein Databank files as three-dimensional molecules into the electronic notebook. As the group analyzed its NMR data, the researchers proposed protein structures that were consistent with the data. At first, the researchers exchanged these files as text in the notebook—as columns of numbers representing the X, Y, and Z coordinates of all the atoms in the molecule. Obviously these numbers are much more difficult to interpret than a three-dimensional rendering of the molecule. Integrating the EMBL software into the notebook made it possible to view the protein structure, rotate the molecule, query bond links and angles, and so
forth, without having to use any other applications. These enhancements make it easy for participants to go from the experiment to the collaboration and back. Collaboration becomes more informal and more integral to the experiment rather than being a separate laborious task that can be delayed or forgotten. Collaboration occurs as part of the research rather than through a report after the experiment is done.
While using appropriate technologies and integrating collaboration tools into the experimental process are important aspects of making collaboratory interactions successful, we also need to consider the effects of these technologies and collaboratory practices on the personnel and organizations involved. For example, today an experimentalist must be a machinist, a pump mechanic, a laser technician, a chemist, and a grant writer to be successful. In collaboratories these jobs can be separated, with individuals specializing in each area and sharing their expertise. For example, in collaborative high-energy physics projects, engineers build advanced light sources and then chemists use those light sources to perform experiments. With collaboratories this type of specialization may occur more frequently, on smaller instruments—at a finer scale—than it does now.
Conversely, collaboratories will allow individuals to expand their range of collaborations more widely as well. Today, chemists who study a molecule and have questions about it that cannot be answered using the instruments available in their own laboratory often have to wait for someone else with the appropriate instrumentation to see the questions in the literature and try to answer them. Researchers have no ability to pursue interesting questions regarding a chemical system through a series of experiments unless they can buy or build all the necessary instruments. Collaboratories will allow researchers to quickly obtain temporary access to additional resources, speeding the pace of scientific progress.
At an institutional level, although one institution might not be able to afford a state-of-the-art spectrometer, four together could obtain one and give everyone equal access to a more capable resource. Institutions can also assemble a scientific ''SWAT team" to tackle a new problem—composed of, say, a geochemist, a chemist, a physicist, and some computer people—without having to build a facility and wrangle participants away from their home institutions before work can begin. Participants can rapidly form an "Institute for the Study of X" by repurposing facilities at six different sites. Such an institute can be exactly what the users want and need.
Collaboratories may prove particularly important in bringing scientists at small undergraduate-only institutions back into mainstream research. While researchers at these institutions would never be able to obtain startup funds to build half-million-dollar-plus instruments to work with undergraduates, collaboratories will allow them to use instruments at other locations. Access to remote peers will be just as important as access to instruments. It can be very difficult to stay current in a two-person chemistry department; having remote colleagues who are easily
accessible over the Internet could help tackle scientific problems, stimulate discussions, and provide an informal way to get peer review comments.
Collaboratories will allow students to participate in experiments more readily than they can today. Undergraduates will be able to join research teams via collaboratories throughout the year, gaining experiences that are available today only through a limited number of short "summer" fellowships. Increased communication between researchers, students, and educators may also help improve the linkage between formal classroom studies and research experiences.
These examples are meant to show that, in addition to deploying collaboration technologies, we will have to rethink the way we do scientific research and change some of our processes, policies, and expectations to realize the promise of collaboratories. The EMSL collaboratory's logo is a mathematical puzzle with three rings that cannot be pulled apart—yet without the third ring, any two of them will slide apart easily. The whole is more than the sum of the parts. The same metaphor applies to collaboratories. They require some technology and sociology to make possible collaboration among specialists in different disciplines. The result is something new and far greater than the individual components.