BLACKSBURG, June 28, 2001 - The merits of linking two fields seemingly as disparate as geographic information systems (GIS) and bioinformatics might not seem obvious, but Virginia Tech's recent symposium linking the two and its roster of renowned participants from both fields raised expectations as well as eyebrows in national technology circles. "Applications of GIS to Bioinformatics" was the first major public forum to cross-pollinate the disciplines, helping to fortify a relatively new, yet highly promising investigative area credited with the potential for enormous breakthroughs.
According to renowned GIS expert Michael Goodchild of UC-Santa Barbara, who presented at the conference, the meeting was long needed. "In GIS and bioinformatics," he said, "there is increasing need to share data among researchers, policy makers, and the general public." Goodchild noted that, more than ever before, the Internet, the World Wide Web, and digital library technology are helping to make this data-sharing possible. "This new collaboration between GIS and bioinformatics," he said, "will allow us to exchange experiences, methods, and best practices in data sharing."
Nearly 80 academic, government, and industry representatives attending the conference from across the country agreed that collaborating makes sense, as GIS and bioinformatics have much in common, most notably digital maps, large databases, and research involving visualization, pattern recognition, and analysis. In general, researchers use GIS techniques and tools to find and track large patterns, for example, geographic distribution of cancer and other diseases in human, animal, and plant populations. Researchers in bioinformatics generally look at very small patterns, such as those in DNA structure that might predispose an organism to developing cancer.
Presenter Geoff Jacquez agreed with Goodchild on the importance of collaboration and data-sharing across the disciplines. Jacquez, who is chairman and chief scientist of TerraSeer, which distributes environmental and health sciences software, alluded to the conference's impact on epidemiology: "The joining of genomics and proteomics with GIS and spatial epidemiology has the potential to provide an enormous breakthrough."
"In the past," Jacquez continued, "GIS and spatial epidemiology focused on finding and monitoring large-scale, population-based occurrences such as disease clusters, outbreaks of infection, or possible associations between cancers and environmental factors. But this exercise was mostly limited to generating hypotheses, not facts. As a result of new dialog between the fields, as we've had at this conference, we are gaining an important mechanistic link between individual-level processes tracked by genomics and proteomics and population-level outcomes tracked by GIS and epidemiology. This will allow us to do a far better job of monitoring, quantifying, and predicting human-health consequences associated with the environment. The potential payoff in related fields such as those looking at climate change, emerging and resurgent infectious diseases, and environmental health is enormous."
James B. Campbell, head of Tech's geography department and an organizer of the conference, said exchanges across the disciplines are mutually beneficial. "Whether the maps used in a discipline are derived from aerial photos taken by a satellite, as with GIS, or discoveries at the microscopic level that are mapped by gene researchers," he said, "many of the processes and technologies are similar or the same; it made sense that we should be talking."
Virginia Bioinformatics Institute (VBI) Director Bruno Sobral agreed. "The notion of a map goes all the way from the level of a genome to a map of the United States," he said. "Bioinformatics has focused on modeling from the level of the molecules up to the whole organism, while GIS has created tools to model from the level of the ecosystem down." Sobral said this makes the individual organism "a perfect meeting point for the two communities."
Interim Provost James R. Bohland, whose office cosponsored the conference with VBI and Tech's Office of GIS and Remote Sensing, called the conference a success. "In terms of attendance, interest, idea exchange, and participants' plans for future collaborations, the conference met and even surpassed our goals," he said. "Given that convening these researchers stands to accelerate research and development in GIS, in bioinformatics, and in areas where they converge ultimately benefiting worldwide populations served by the outcomes we were very pleased to push this dialog."
According to Sobral, the idea for the conference was hatched when Bohland, whose urban affairs and planning research taps the tools and techniques of GIS, completed a tour of the bioinformatics institute. "We realized," Sobral said, "that for a university with a major bioinformatics thrust, a significant number of faculty in five colleges with GIS-related research, and an environment known as a hotbed for interdisciplinary explorations, a conference on this topic that was a first-ever in terms of size, scope, and attendance was a natural."
The idea of connecting geneticists with GIS specialists could just as easily have come from researchers' listening to themselves talk, for the seeds of collaboration have long been embedded in the lexicon and metaphors of both fields.
Genomics participants' language was ripe with references to genetics stated in geospatial terms "deserts" of uncharted genetic space, "neighborhoods" of genes, and genetic "pathways," for example. When a GIS researcher borrowed one of Sobral's slides defining bioinformatics and noted that most of the text still applied when "GIS" was exchanged for "bioinformatics," it was clear that the fields define their needs in similar ways.
Sobral, himself an internationally recognized bioinformatics researcher, said that the conference was bolstered by participation from two internationally recognized GIS researchers: Goodchild and Barbara Buttenfield, who is from the University of Colorado. "We also heard from Carol Bult of the University of Maine, who created the first direct application of GIS technology to bioinformatics and hosted the first spatial genomics workshop during 1997," said Sobral. The work of Stephanie Green from Washington State, John Havlin from North Carolina State, and Tech panelists Bill Carstensen (geography) and Randy Wynne (forestry) was also featured.
Broad interplay between geography and biology isn't entirely new. Software companies such as GIS giant ESRI and Jacquez's TerraSeer have for some time marketed software for health and natural resources applications. Representatives of both companies were in attendance, however, to survey some of academe's first steps into a new area. They also shared perspectives from the marketplace on GIS's future in bioinformatics and on applications of GIS to environmental epidemiology and ecology.
Jacquez described what was, from his perspective, the symposium's biggest take-home lesson: "After years of using GIS data to track diseases in populations in terms of the what, where, and when, the integration of bioinformatics data, genomics and proteomics data telling the story of what takes place at the cell and sub-cell level in individual diseased organisms will soon enable epidemiologists using GIS to capture the how of disease outbreaks." Integrating bioinformatics with GIS, he said, will be "phenomenally useful" in predicting public health outcomes.
Encouraging dialog among the fields, Sobral said, will enable researchers using GIS, for instance epidemiologists, to home in on individual cases. At the same time, he said, geneticists will be able to widen their purview, "noting, for example, how the incidence of congenital disease varies over space, over time, and with changes to terrain and climate."
Shared technical challenges and applications relevant to both disciplines were other topics for discussion. Both fields rely heavily on mining, managing, accessing, and analyzing large amounts of data, Bult said, and the disciplines share many of the same challenges for data management and representation. "We learned that many solutions to technical problems arrived at in one discipline can serve as object lessons for the other." Bult added that when some of the display and interactivity features of GIS maps are ported to genomics maps, genetic maps will go from being mere display tools to serving as interactive "mechanisms for discovery."
Campbell said other possible long-term outcomes of connecting the two investigative fields are better modeling of the spread of insect infestations, for instance that of the gypsy moth in the United States; better monitoring of diseases such as rabies in wildlife populations; improvements in agricultural productivity; and reductions in the negative environmental impacts of some agricultural activities. "The cross-pollination of research accomplished by the symposium," he said, "will accelerate multiple outcomes not even yet envisioned."
For a limited time, proceedings, which consist of streaming video files and associated slide presentations, are available at http://www.vbs.vt.edu/.
June 27, 2001