Peter Hansen and Petra Krauledat remember it well. The husband-and-wife team had just built a machine for a small Cambridge company to study the creatures. And they had lost a lot of money on the contract.

Then, suddenly, this obscure little worm was propelled to international scientific celebrity. Its genetic makeup, it turned out, was remarkably similar to that of humans. It could help scientists understand the role of certain genes and develop drugs to treat diseases such as Alzheimer's and diabetes.

"In 1998, nobody had even heard of this worm other than some nerdy biologists," said Petra Krauledat, a biologist herself and president of Union Biometrica Inc. "By mid-1999, every drug company had a program using these nematodes."

And the machine that Hansen and Krauledat had considered a financial loss had suddenly become the key to unlocking the nematode's potential -- and their company's.

The machine could do in a few weeks what took scientists peering through microscopes more than a year to do. It could test tens of thousands of drug candidates on hundreds of thousands of living nematodes, identifying those that work.

And, eventually, Hansen and Krauledat said, it could drastically shrink the time, risks and cost of drug discovery for humans.

"One of the lessons of the human genome project is that automation is essential to basic life-science research," said Hansen, a physicist and Union Biometrica's chairman and chief technology officer. "The day of the white- coated laboratory researcher is gone."

Drug discovery is an expensive and risky business. Out of 10,000 promising compounds, only one will make it to market. It will take an average of 15 years and $300 million to get there. And more than half of that money will be spent before the drug even reaches human clinical trials.

The very survival of pharmaceutical and biotechnology companies depends on improving those odds. As medicine advances, drug companies are tackling more complex and elusive diseases. They are pouring more money into research, meaning they must also deliver more drugs to stay financially healthy.

At the same time, the public outcry over drug pricing is increasing, forcing companies to explore more cost-effective and less risky ways of developing drugs. The crush of data streaming out of genomic research is also demanding faster and more systematic approaches to drug discovery.

Faced with these challenges, drug and biotech companies are turning to robotics, optics and high-performance computing to speed the pace of drug discovery.

Robots stack, move and place trays of cells onto machines that record the cells' reactions to various doses of drugs. Computers systematically match libraries of drug compounds with models of genes. Huge databases of genetic information scan other databases for similar genes.

But the automation typically stops when drug compounds leave the research bench to be tested on mice or other animals. Scientists genetically alter the mice to mimic human diseases, give them the test drug, observe them and later autopsy the mice to determine if the drug had any effect.

The process is time-consuming, labor-intensive and costly. It can take months for researchers to study a reaction in the mice. Often, the discoveries are made only after killing and dissecting the mice to study their biological responses to the disease and the drug.

That's why pharmaceutical companies and the scientific community greeted with open arms the surprise discovery that nematodes and humans were genetically similar.

Compared to mice, nematodes are cheap and easy to study. The worms are transparent, allowing researchers to track a gene inserted into a nematode under a microscope by adding a fluorescent protein to the gene. And if a drug turns off the gene -- a sign that the drug is working -- the glow disappears.

The worms also reproduce quickly, making hundreds of exact copies of themselves in a few days. And since they are tiny (one-millimeter long as adults), they can be easily stored in small petri dishes or test tubes.

But there are some drawbacks. Their size and tendency to wiggle make them tricky to catch. Scientists peer through microscopes, grabbing the worms with a sliver of platinum wire. The work is tedious and tiresome. And although faster than testing on mice, it takes days to analyze 1,000 nematodes.

By automating the process, however, Hansen and Krauledat have shrunk days' worth of work into a single minute. And in doing so, they invented the only automated system of testing living organisms. The machine also can be adjusted to test other small, simple organisms such as fruit flies and zebra fish.

Called the COPAS Technology Platform, the automated testing system feeds the creatures one by one into a tube and through a scanning system that reads the fluorescent genes. The more the worm or fruit fly glows, the more active the disease.

The scanner relays the information to the computer, which plots the worm on an on-screen image according to its age and level of disease. The computer then instructs the machine to sort the worm accordingly.

The COPAS system does all this at the rate of 100 nematodes per second. In a rapid-fire fashion, researchers can test thousands of drug compounds and spot the ones that dim the glowing genes. They can also weed out compounds that seem to have toxic side effects.

Historically, when researchers have begun testing a drug on a mouse, they've known very little about the drug. It looks promising when tested against tissue samples or single cells in a test tube or petri dish. But they have no information about how it might work in a living organism.

Testing on nematodes or other simple organisms can help narrow down the candidates that move through the pipeline, reducing the risks and costs of failing later in the process.

The testing machine costs between $200,000 and $700,000, depending on the complexity of the system. So far, companies that have bought it are using it as an intermediate testing stop between the lab and the mouse.

As the technology spreads, however, Hansen and Krauledat said it could be used to replace lab tests on cells and tissue samples, making testing on nematodes and other simple creatures the first and last step before testing on mice.

On average, Krauledat said, it now costs a company $200 million to get a drug from the idea stage to the first human clinical trial. Hansen and Krauledat said their technology can shrink that number to $50 million over the next five years.

Hansen and Krauledat said they haven't advertised the machine or published their findings in scientific journals. Indeed, references to Union Biometrica are few and far between in the mainstream media. They've been too busy, they say, to worry about publicity.

Word of mouth, however, has brought some of the world's largest pharmaceutical companies to their door. Among their customers are industry giants like AstraZeneca PLC, Aventis SA, Johnson & Johnson, Novartis AG, SmithKline Beecham, and Glaxo Wellcome PLC.

In May, after a decade of supporting the private company on revenue from their contract work, Hansen and Krauledat accepted $2 million in venture capital. It was time, they decided, to concentrate on developing and marketing their new technology without the distraction of scrambling for cash.

The Somerville, Mass., company will finish this year with $4 million in sales, a figure Hansen and Krauledat expect to double next year. With a backlog of 130 labs interested in the machine, they said, their biggest problem is hiring people to meet the demand.

"To find out what all these newly discovered genes do and how to make drugs that affect them is going to require new kinds of equipment and new ways of doing things," Hansen said. "That's what we're really offering -- new ways to study disease and discover drugs to treat those diseases."