Analysis of that data required 460 days of computer processing time, but it was condensed into a few short weeks by use of parallel computing, Vision said.
Currently, scientists know the complete genetic makeup of only a handful of plants although humans are economically dependent on dozens of different ones, he said. It would be far too expensive and time-consuming, however, to characterize the genomes of all of them experimentally.
"Many plants are impractical to work with experimentally because they have genomes full of material that serves no known function," Vision said. "For example, the lily genome is roughly 40 times the size of the human genome."
Nevertheless, the gene content of lilies can be predicted reasonably well by looking at a related plant with a small genome, like rice, he said. To breed better crops, it's important to be able to leverage the information from the tractable model systems that scientists already know so much about.
"We will continue to add new features to Phytome over the next few years," Vision said. "Perhaps the most important will be the ability to compare the genetic maps of multiple species simultaneously and predict the gene content in regions of plant genomes that have not yet been deciphered."
Besides Vision, others involved in the development of Phytome are Dr. Stefanie Hartmann, a postdoctoral researcher in biology; Dihui Lu, a graduate student in information and library science; and computer programmer Jason Phillips.
The National Science Foundation is supporting the project with a five-year, $1 million grant it awarded to UNC in 2002.
"This is a unique resource for scientists trying to understand the genes contributing to variation in traits of economic importance in crops," Vision said.
Contact: David Williamson
University of North Carolina at Chapel Hill