By Mark D. Uehling
July 11, 2002 | You’re a farmer in Mississippi. Tobacco budworms, a.k.a. Heliothis virescens, are devouring thousands of acres of your tobacco, your cotton, your corn. The bugs are increasingly resistant to conventional chemicals -- even cotton that has been genetically engineered to repel insects. Your next defense? Data.
That’s the word from Exelixis Inc. of South San Francisco. Late in the spring, the company surprised biologists with the news it -- and a partner, Bayer AG -- had sequenced 90 percent of the roughly 15,000 genes of the tobacco budworm.
The new sequence is the first known decoding of a member of Lepidoptera, the family of fluttering things -- butterflies and moths. “This family causes 40 percent of the insect damage to crops around the world,” says Klaus Raming, general manager of GenOptera LLC, the 8-year, $80-million research venture between Bayer and Exelixis. The moth is hardly obscure in agricultural circles: It eats hundreds of millions of dollars worth of crops annually around the world.
A Secret Sequence
For all its importance, the new tobacco budworm sequence will not be deposited in a public database for all scientists to use. In academia, that is raising a few eyebrows. “I would hope they would share with the public so we can work on this together,” says Linda J. Gahan, assistant professor of biological sciences at Clemson University and a leading budworm researcher.
GenOptera, for its part, says it is willing to work with selected academic researchers, and to share limited amounts of its sequence. The most pressing task, however, is using its data to accelerate the development of pesticides. “This approach enables us to get totally new targets,” says Raming.
Those targets are biological chinks in the armor of the insect -- proteins with which pesticides could interact in a no-nonsense, life-ending manner. The GenOptera team in California is supplying those molecules, which are being further analyzed for safety and lethality in Germany. Because the tobacco budworm is closely related to other hungry bugs, the work could lead to chemicals that affect more than just the budworm.
One objective in picking the targets is to find proteins as unique to insects as possible. Methodically selecting proteins limited to the insect world was unthinkable in the era of conventional pesticide research. If it works, that approach could be better for the environment and the companies’ bottom line.
Biology on a Budget
These days, of course, simply sequencing a new species will not get you shortlisted for a Nobel Prize. Anyone with $70 million or $100 million can commission the genetic sequence of a favorite pet or a creature not yet unraveled by molecular biologists.
What makes the Exelixis research interesting to bench scientists and IT types is the way it was done -- quickly, cheaply, and with an emphasis on practical results. Many academic sequencing projects are designed with the lofty scientific objective of finding 100 percent of a species’ genes, no matter how long it takes or what the cost.
That was unappetizing to Exelixis and Bayer. “If you’re a company,” says Darren Platt, head of the computational target discovery group at Exelixis, “you really want to ask the question differently. You want to ask, ‘How much money do I need to spend to get the answer I need?’”
Platt continues, ticking off a few numbers. “You could spend $10 million or $20 million to get a complete insect,” notes Platt. “But if there are 10 or 12 insects out there that you might want to look at, you might want to start sampling them. The fact that these genomes are larger is one of the factors driving more efficient sequencing methods.”
His IT marching orders, it turns out, originated with life science colleagues. As Platt explains: “What our researchers said was, ‘If you could give us 80 to 90 percent of Heliothis, for much less cost than it cost to do the whole job of sequencing every last base pair, that would be all we would need.’ That was the computing challenge. Was there a strategy that would get you 80 to 90 percent of the sequence at, say, a twentieth of the cost?”
To answer that, Platt turned to proprietary Exelixis software called SimSam. (“Sim” stands for simulation, “sam” for sampling.) The program is written in Python and runs on a four-processor Sun Solaris box. “We simulate the genome,” says Platt. “We build a fictitious, virtual Heliothis.” The program can generate an ersatz genome overnight.
The genome-building software allowed other informatics tools to be tested, Platt says. “You can feed the genome into all of the tools you’re going to use to look at the real genome,” he notes. “It’s a very good test of ‘can you handle the data when it comes along?’”
With an approximation of Heliothis, Exelixis could plot an effective sequencing strategy that combined expressed sequence tag and shotgun techniques. The computer predicted that strategy would provide the results the bench scientists had requested at a tenth the time and cost of finding all of the tobacco budworm’s DNA.
“The genome is not actually the product,” Platt says patiently, noting that novel molecular targets, or attack points for new pesticides, were the real objective. Winnowing down the vast number of potentially interesting genes, in turn, was expedited by Exelixis’ body of knowledge of the fruit fly, whose genes could quickly be compared to tobacco budworm genes. So the cloning of potential targets for new pesticides began even before the sequencing was complete.
While Exelixis uses the standard open-source tools to compare snippets of genetic material, its strength in comparative genomics also depends on in-house expertise in computer visualization. Like all its tools, this software module is available to Exelixis partners, which include Dow AgroSciences and Aventis CropScience, not to mention Bristol-Myers Squibb, Merck, Schering-Plough Research Institute, Cytokinetics, and Protein Design Labs.
“The thing that is more important than having the tools is being able to view the results,” Platt says. “We have proprietary ways of visualizing the data and filtering it. If you have to look through 100 genes, you’d do it one way. If you have to look at 1,000, you’ve really got to train the computer to be pretty smart, to pick out the things that a human being would notice.”
Scientists at Exelixis, meanwhile, are already working to bring its research full circle, applying insights from Heliothis to pharmaceutical research. “The chairman of our board says flies are little people with wings,” says Lloyd Kunimoto, CEO of agronomics at Exelixis. “You can study a lot in the fly that tells you a lot about how to develop drugs.”
“One of the things we discover when we turn genes on and off is which genes are essential for life,” Kunimoto says. “It was not a great intellectual leap to say maybe we can use knowledge of which genes are essential to develop new ways of killing insects.”
Smoking Out The Tobacco Budworm
- U.S. acres infested with budworm and bollworm: 11.5 million
- U.S. cotton crop losses to budworm/bollworm in 2000: $300 million
- Bayer employees: 117,000
- Exelixis employees: 570
- Bayer revenues in 2001: $26 billion
- Exelixis revenues in 2001: $41 million