HINXTON, UK--Integrating state-of-the-art graphics and visualization techniques into bioinformatics research will be very beneficial, if not of paramount importance, to make sense of the enormous data influx that is predicted for the next decade into this area of the life sciences. The Frontiers in Visualisation and Human-Computer Interaction Conference '97, held in the Conference Centre of the Wellcome Trust Genome Campus here December 10-11, made a significant first step toward bringing researchers from the two worlds together. The event was organized by the European Bioinformatics Institute (EBI).
"The conference aimed to provide an arena for scientists and professionals in the biosciences and pharmacology to meet with the leading graphics and interaction experts and to develop new ideas and synergies," explained Chris Jones, a coorganizer of the conference along with EBI's Alan Robinson. A researcher with the CERN research center in Switzerland, Jones is currently on sabbatical at EBI. "The conference was attended by 250 people, with about a fifth of the participants originating from the graphics or other nonlife-sciences professions," he observed.
The first day of the conference was intended to give scientists an overview of the state of the art in graphics and human-computer interaction. Graphical widgets, virtual and "augmented" reality, caves, real-time collaboration tools, and several props such as force-feedback joysticks, power-gloves, and belt computers were only some of the techniques and utilities shown.
Among speakers, Brown University's Andries van Dam, director of the US National Science Foundation's Science and Technology Center for Graphics and Visualization, addressed the move away from traditional input devices that have dominated and limited interaction with applications heretofore. To demonstrate he showed three-dimensional interaction widgets and applications for scientific visualization, controlled by mice or other interaction devices with three or more degrees of freedom.
One particularly striking example was the NanoManipulator, a combined display table and force-feedback pen that allowed a researcher to examine an object. In this case, several microscopic slides through a cell body were used to provide a three-dimensional structure and the researcher could use the pen to "feel" the texture of the object.
José Luis Encarnação, of the Technical University and director of the Fraunhofer Institute for Computer Graphics in Darmstadt, demonstrated how scientists face a shift from "presentation and interaction" to "navigation and experimentation." Supported by new presentation technologies, data types, interaction approaches, teletechnologies, and teleapplications, the trend will enable visual information in computer-generated environments, opening new opportunities for applications in experimental sciences and for applications that have to be executed independent of time and location.
The program included several live demonstrations by Detlef Krömker, of the Fraunhofer Institute of Computer Graphics, of techniques and applications that were discussed by other speakers. The demonstration of a projected virtual environment for interactive molecular modeling was a highlight. Using a head-mounted display (the audience received 3-D glasses) and data gloves a scientist manipulated the positions of an AIDS antiviral drug and reverse transcriptase enzyme with respect to each other while a background molecular dynamics simulation gave feedback on the feasibility of the interaction.
Jean-Francis Balaguer of Artemedia Online presented features and advantages of Virtual Reality Markup Language (VRML) and Java 3-D for use on the Inter net. However, the audience commented that there is a reluctance on the part of major players such as Microsoft and Silicon Graphics to commit to Java 3-D, especially because there seem to be few content applications at the moment. One thing bioinformatics can do to help encourage Java 3-D is to supply interesting applications, Balaguer noted.
On the second day the conference focused on the application of visualization to the biosciences. The first speaker was Christian Henn, manager of Silicon Graphics' European Visionarium, with a presentation on smart Multi-User Virtual Environments, which can be used as real-time collaborative environments to visualize complex technical and scientific information and to allow scientists to work together on a problem. Henn demonstrated the concept by working with a colleague at another site on a three-dimensional rendering of a protein. The synchronized views of the browsers at either end showed how well the technology already operates.
"Everything that links to data abstraction or abstract visualization of information is applicable immediately to all sorts of bioinformatics research. In addition to that, bioinformatics information is always going to be linked to three-dimensional structural information, and that is the ultimate goal," he commented. "The more structural information we have and the more we understand the relationship between structure and function and sequence, the more we will see that all these things merge together." A similar approach based on work of Phil Bourne and associates at the San Diego Super computer Center, focusing on macromolecular structures, described a Molecular Scene Description Language that can be used to store scenes and additional information in a relational database, ready to be queried and rendered in a VRML browser to enable collaboration between scientists over the Internet.
Howard Bilofsky, head of advanced information technology for SmithKline Beecham, stressed how scientists today are confronted with the paradox of having both information overload (in proprietary pharmaceutical databases and remote sources such as the Internet) and starvation (not being able to find required information quickly, if at all). Intelligent integration of information based on mediator architecture combined with novel visualization techniques will help bring critical pieces of information to the surface, Bilofsky remarked.
In his talk, Robinson discussed work on visualization widgets and described how they easily combine to make powerful analysis tools for bioinformatics. John Boyle, of Robert Gordon University, did the same for the Java visualization beans developed in the area of cancer chemotherapy. Other speakers included John McLachlan of the University of St. Andrews, who discussed his work on virtual reality three-dimensional modeling of embryo structures, and Gareth King of Zeneca Pharmaceuticals, who addressed issues in analyzing huge amounts of microarray and large-scale cDNA sequencing data on gene expression. Visualization and data-mining tools such as Silicon Graphics' MineSet and the in-house-developed GeneVis were successfully used to tackle this challenge, said King.
A lively panel discussion concluded the conference. One concern was whether high-end visualization techniques will really help bioinformatics in the short-term. Encarnação and van Dam argued that while the techniques are now focused on other fields, cooperation with bioinformaticians is required to help develop tools that can make the difference "between winner and loser in the pharmaceutical arena." They strongly stressed the need for interdisciplinary teams to organize work flows within bioinformatics or pharmaceutical research and make prototypes. Also, they emphasized the need to influence software vendors and not be content with waiting and hoping that you end up with what you want. Paolo Zanella, head of EBI, added that the swing toward making visualization a priority will come when companies or individual scientists openly attribute their success to the use of visualization tools.
After the conference, Bilofsky commented further on concerns raised during the panel. "I think we just scratched the surface of how to apply some of these very sophisticated computer science and visualization techniques in the area of bioinformatics, and bioinformatics itself being such a new discipline, I expect to see enormous synergy between the two," he remarked. "The problems we have in bioinformatics will in fact encourage the computer technologists to change and improve the tools they provide."
Henn added, "Events like this are really good steps in bringing the visualization and bioinformatics camps together, and in fact I think there are already some initiatives being started. Quadstone's and EBI's joint BioVis project is going in that direction."
There was a strong feeling among participants that the conference should be repeated, and that supplementary conferences and workshops focusing on more detailed problems and solutions are needed to bring bioinformatics quickly up to speed with the newest visualization techniques and technologies.
--Jean-Jack M. Riethoven