Deep within a plant cell lies the code that controls its form and function. Steven D. Tanksley, Cornell's Liberty Hyde Bailey Professor of Plant Breeding, and Susan R. McCouch, Cornell assistant professor of plant breeding, want to crack this code and unleash its potential.
To perform this code-breaking, the two researchers are employing high-performance supercomputers at the Cornell Theory Center (CTC), focusing on the genomes of Arabidopsis, rice and tomatoes. Tanksley and McCouch hope that, with the help of CTC's computing resources, one day they will be able to provide a legion of plant breeders and genetic engineers with a road map of the genetic codes of important crops.
Genetic codes are vital information to plant breeders who seek to increase the yields of specific field crops. Domesticated rice, for example, has reached its yield potential around the world. Tanksley and McCouch aim to show that yield genes from wild species of rice can be recombined to bring a new, more productive life to popular rice varieties.
Genetic sequences and their functions in major crops are being identified with increasing speed. But simply mapping the genes is not enough. Tanksley and McCouch are running a program that allows CTC's IBM SP parallel supercomputer to search for genomic landmarks, which are bits of knowledge about gene structure and function. The program then compares these genetic patterns in a particular hybrid or species of a crop with those of its relatives. The patterns are then archived on a genome map.
The program software, appropriately named BLAST (an acronym for Basic Local Alignment Search Tool), enables the supercomputer to look for these genomic landmarks in one one-hundredth the time it would take a high-speed workstation computer. Much of the research in computational genomics is done with parallel computers -- harnessing computers together much like a team of horses.
The basis for computational genomics consists of computer science, structural biology, molecular genetics, statistics and applied mathematics, says McCouch. Such an interdisciplinary approach is critical when searching entire genomes. And with an eye to training future researchers in this field, McCouch and her colleagues in the plant breeding department have organized Cornell's first biology course on computational genomics.
In addition to its IBM supercomputer, CTC is testing an Intel NT system for the mapping of genomes. The new system is part of Intel's Technology for Education 2000 project at Cornell. The U.S. Department of Agriculture also is funding the purchase of computing resources that will be dedicated to the plant genome project.