Heroes of Manufacturing These innovators sail against the prevailing winds, discovering whole new worlds in biotech and software.
By Gene Bylinsky

(FORTUNE Magazine) – Lee Hood: the man who automated biology

It took thousands of years for manufacturing to advance from the ancient Egyptians' primitive, hand-operated lathes to today's lightning-fast computer-controlled production machines. But it took just three decades to automate modern biology, thanks to Leroy E. Hood. Hood, a 64-year-old biologist-technologist from the plains of Montana, is the brains behind four extraordinary biotech machines. His devices include protein sequencers and synthesizers, which open up crucial insights into the structure and functions of proteins, and "gene machines," which read and assemble the chemical components of DNA and have made possible the rapid deciphering of the human genome. "Without Lee Hood and his inspiration for a DNA-sequencing machine, the Human Genome Project would today still be in its infancy," says James D. Watson, the co-discoverer of the helical structure of the DNA molecule.

Hood's machines have also had profound medical--and commercial--impact. They have enabled Stanley Prusiner to pinpoint a misfolded protein as the cause of mad-cow disease and its human counterpart, and helped Hood and his team locate abnormal genes that cause cancer. They've also led to the development of Epogen, biotechnology's first billion-dollar drug, as well as other big biotech medicines. In an industry that often seems dominated by outsized egos, Hood comes across as unusually self-effacing. Despite his MD and Ph.D. in biochemistry, he talks about the impact of his new technology in terms any manufacturing shop steward would recognize: "Higher throughput, lower costs, high quality, high quality control." In typically understated fashion he describes what he does as "manufacturing good data."

Prosaic-sounding, perhaps, but the good data manufactured by Hood's machines have resulted in a Nobel Prize for Prusiner, and Hood could well be on his way to a Nobel of his own. The most recent recognition of his achievements came late last year when he won the $410,000 Kyoto Prize for advanced technology. As part of the festivities, Hood spent 45 minutes with the Japanese Emperor and Empress in the Imperial Palace.

Tokyo is a long way from Hood's childhood homes in Missoula and Shelby, Mont., where he was one of four children of a phone-company engineering executive and his wife, a high-school teacher. Encouraged by his parents to camp out and ride horses when he was as young as 5, Hood developed a fierce spirit of independence and self-reliance. He was drawn to biology early on, inspired in part by a brother's Down syndrome, and became so proficient in the subject that as a high-school senior he taught it to sophomores. Under his father's guidance, he also mastered the intricacies of electronic circuits. And well-rounded? Hood was his football team's quarterback, leading the squad to a three-year undefeated run.

Hood's high-school years were also heady times in biology. Watson and Francis Crick had discovered the structure of DNA in 1953, and Hood was excited by the tremendous potential of the field. In 1956 he landed at that hothouse of innovation in physics and biology, the California Institute of Technology in Pasadena, where he studied under the likes of Linus Pauling and Richard Feynman. But he fretted that he didn't know enough about how the human body functions--so he interrupted his Caltech studies to earn an MD in three years at Johns Hopkins University.

Early on, Hood recognized the deep insights that could be gained by analyzing the building blocks of life. After returning to Caltech to pursue his Ph.D., he learned protein "sequencing," which is technical jargon for determining the order of the 20 amino acids that make up a protein, and thus figuring out what the protein is. Proteins are like biological micromachines made by genes to execute life processes; identify and analyze the relevant proteins, and you have a shot at understanding fundamental biological activities, such as the body's defense against infection. Working with his professor, Hood showed that antibody proteins were encoded by two separate genes, a finding that helped explain the immune system's versatility. The idea flew in the face of the prevailing belief that a gene could make only one protein, and met with skepticism and even scorn. "I realized for the first time how threatening new ideas are to many scientists," Hood says.

He also grew impatient to speed up the process that generated such ideas, and in the late 1970s began working to develop a machine that would reliably and rapidly automate protein sequencing. But a protein sequencer presented a plumber's nightmare of beakers, valves, a spinning cup, a scanning laser, and other components. The basic idea was to use various solvents and reagents to tag a protein's individual amino acids and then read the results. To make the machine work, Hood and his team spent years designing nonleaking valves capable of handling the highly corrosive solvents. When a working prototype finally emerged in 1980, it could function with quantities of protein 100 times smaller than a primitive Swedish predecessor required. Because it allowed scientists to analyze proteins impossible to isolate in large amounts, Hood's protein sequencer literally transformed biology. Hood and his team, for instance, used it to discover that cancer genes in many cases are normal growth-control genes whose code contains "misprints."

In another major feat in 1982, Hood and his associates used their sequencer to determine the structure of the hormone erythropoietin, which stimulates the production of red blood cells. That discovery led to the formation of biotech's most successful company, Amgen, of which Hood is a co-founder. Amgen's version of the hormone, Epogen, is used to treat anemia in dialy-sis patients, and it has hit total sales of $15 billion since its introduction in 1989. (Hood helped start ten other biotech companies, including Systemix, Darwin, Rosetta, and MacroGenics; though comfortable, he says he isn't wealthy, having turned over the royalties on his machines to his lab and his company stock to an educational foundation.) The sequencing of other substances, such as interferons and colony-stimulating factors, followed, resulting in more big biotech drugs.

Other labs understandably clamored to have Hood's sequencer. Caltech wasn't interested in commercializing the machine, though, and neither were 19 companies that Hood visited. So in 1981 he used $2 million in venture capital to found Applied Biosystems, which would eventually build and sell all four of his machines. (It's now part of Applera Corp. and had annual revenues of $1.6 billion in fiscal 2001.)

Hood's second machine, the protein synthesizer, came in 1984. The device made it possible to produce long proteins in high, consistent yields so investigators could run experiments on them. Together with scientists from Merck & Co., a Hood associate used the machine to synthesize and purify a protein that became the basis of a protease inhibitor, one of the most effective anti-HIV drugs.

At the same time Hood and his colleagues developed a DNA synthesizer, which puts together fragments of genes. This machine made it practical to string together genes for use in sequencing and cloning. Combined with the other new machines from Hood's lab, the synthesizer would make possible yet more discoveries, such as pinpointing the cause of mad-cow disease. The toughest was the gene sequencer. Hood and his team were attempting the "impossible," said a National Institutes of Health (NIH) panel, which turned down his request for a grant to help build the instrument. Conventional lab scientists scoffed that graduate students could decipher genes by hand a lot more cheaply. With typical determination, Hood persisted. "Lee is a true visionary who saw opportunity in a place where conventional wisdom would deny its presence," says Caltech president David Baltimore.

The technical challenges were formidable. Strands of DNA consist of an alphabet of four chemicals in different combinations. In manual gene sequencing, researchers run four different chemical reactions to sort DNA fragments based on this alphabet, then use several subsequent steps to label, sort, and detect the fragments. It made for a time-consuming and not always accurate process. Hood and his colleagues hit upon the idea of labeling DNA fragments with fluorescent dyes, using a laser to make the DNA chemicals glow in red, green, blue, or orange, and then reading those signals by computer.

Sounds simple enough. It just took a team of Caltech biologists, chemists, computer and electronics engineers, physicists, mathematicians, software experts and others six years to work out the details. The DNA sequencer emerged in 1986 to become the workhorse machine of the Human Genome Project, which completed the reading of the basic blueprint for human life in 2001, five years ahead of schedule. Baltimore calls the gene sequencer the "greatest" of Hood's machines.

Although he had become chairman of Caltech's biology department in 1980, Hood was still looked upon by old-style biologists as a brash interloper. They much preferred their traditional small-team hypothesis-testing approach. Here was Hood pushing his "discovery science"--the idea that technology can advance biology by large leaps--and assembling huge teams with flashy new production machines and computers that threatened to take over old-fashioned biology. While Hood doesn't come across as an iconoclast, he does have "strong feelings about what he thinks is the pathfinding science of the moment," says Baltimore.

Unable to convince his colleagues that Caltech should back his approach with the creation of a new department of molecular biotechnology, Hood left for the University of Washington at Seattle in 1992. After a four-hour dinner, one fascinated local, Microsoft's Bill Gates, kicked in $12 million to help start Hood's department. Just as at Caltech, machines and ideas poured from Hood's new lab, among them an instrument that uses inkjet-printer technology to synthesize large arrays of DNA on glass slides, where patterns can be examined for the errors that cause disease.

And in 2000, Hood launched yet another new venture. For years he has been fascinated by systems biology. That's a school of thought that advocates studying all the complex interac-tions in a living system simultaneously rather than scrutinizing one gene or one protein at a time, as biologists have done for decades. Systems biology also calls for the integration of biology, medicine, computation, and technology, an approach Hood believes holds the secret to understanding diseases at the molecular level.

So Hood and two colleagues have started the nonprofit Institute for Systems Biology in Seattle. Supported by federal grants and industrial collaborations with companies like IBM, Merck, and others, as well as by private donations, the institute already employs 170 scientists, engineers, and technicians. Some important findings--and, of course, new machines--are already emanating from Hood's latest pioneering venture.

Dan Dodge and Gordon Bell: When software really, really has to work

Is software hopeless? Ask anyone whose computer has just confessed to an illegal operation or whose screen has locked up. Despite decades of effort, a wisecrack from the software industry's early days still stings: "If builders constructed buildings the way programmers write software, the first woodpecker to come along would cause the collapse of civilization."

There's one notable exception. As far as anyone can tell, software created by a Canadian company called QNX Software Systems simply doesn't crash. QNX's software has run nonstop without mishaps at some customer sites since it was installed more than a decade ago. As a delighted user has put it, "The only way to make this software malfunction is to fire a bullet into the computer running it."

Like Windows or Linux, QNX's program is an operating system, the traffic cop that organizes and runs a computer's many functions. But this operating system is used mostly in highly specialized, real-time industrial applications. QNX software directs "extreme" manufacturing, such as guiding the flawless grinding of optical lenses--a process in which the slightest software glitch can ruin a product worth $100,000. It's also used to control facilities such as nuclear power plants and other critical installations where any software funny business could be catastrophic.

QNX's software is the brainchild of Dan Dodge, 48, the company's CEO, and Gordon Bell, 47, its president (they swap titles every year). They're subdued, low-key fellows--until you ask about their technology. Then they grow animated and even passionate. They're friends who drive basic cars no different from those of their employees and live in modest houses. Tucked away in a nondescript industrial park in Kanata, Ontario, an Ottawa suburb 2,400 miles from the hubbub of Silicon Valley, QNX has been called a "stealth company," its founders say.

It may not be on many radar screens, but it's hardly stealthy to smart manufacturing and process-control engineers, who have installed more than one million QNX systems around the world--and beyond. Cisco uses QNX to power some of its network equipment; Siemens uses it to run its medical systems. QNX enables the high-speed French TGV passenger trains to round curves without tilting too far; it runs U.S. Postal Service mail-sorting machines, directs GE-built locomotives, and will soon control all of New York City's traffic lights. The Federal Aviation Administration (FAA) has purchased 250 copies of QNX for air-traffic control. And the system operates the Canadian-built robotic arm on both the space shuttle and the international space station.

From the earliest days, Dodge and Bell chose to focus on such industrial-strength applications for their software. But QNX was designed to run on PCs, and its latest version, Neutrino, features an easy-to-use graphical interface, e-mail, web browsers, and the like. The company says tens of thousands of users, mostly computer whizzes, have downloaded a free version of Neutrino from its website. And QNX software may soon become known to more than just plant managers and aficionados. Versions of the ultra-reliable system are beginning to push out Microsoft and other competitors in new, smaller-scale applications. QNX powers some new laser eye-surgery devices, portable home-dialysis machines, and computerized casino games, for example.

Some car owners will soon be steering to directions from QNX. IBM has chosen the software for an advanced automobile-navigation system that integrates traffic updates, weather data, and emergency service information on the fly. The system, which automatically re-computes routes to bypass accidents, traffic jams, or dangerous road conditions, will appear in luxury models in the U.S. in the next year or so, and more broadly soon after. "We'll be shipping 20 million systems to the automakers in 2004," Dodge says. QNX expects to dominate the software part of this new market, which UBS Warburg predicts will be worth $9.3 billion by 2005 (with software alone accounting for 25% of that total).

Dodge and Bell both grew up engrossed by electronics. The son of a Nortel engineer and a housewife, Dodge built circuitboards and even a laser in high school, and in 12th grade he was programming the mainframe at the community college in Belleville, Ontario. To understand electronics, he majored in physics in college but continued to spend most of his time in the computer lab.

Bell was the oldest of four children of a Toronto electronics technician and a secretary at an R&D firm. His father worked for a chipmaker, so Bell "got tons of electronic parts to play with," he says. In high school he was already building computer peripherals such as tape readers and audio systems, and he had built his first computer by the time he went to college--winning one of three provincial scholarships for high-school superachievers. Dodge and Bell met at the University of Waterloo near Toronto, when a friend who knew that both built computers in their dorm rooms introduced them. They've been friends ever since.

Dodge and Bell were long entranced by the idea of creating an operating system that would work without failures. Most programmers have organized their operating systems in the so-called monolithic fashion, in which both the "kernel," or the code that directs the OS, and the various internal computer functions such as file systems and graphic displays, are all assigned a common memory space. The problem with this approach, still followed in widely used operating systems such as Windows and Linux, is that a trivial error in any one of the many functions that share the same memory space can shut down the whole system. The monolithic system, as Dodge and Bell describe it, is like an orchestra in which the conductor, or kernel, and all the players, or functions, are confined to a single room where any problem quickly becomes contagious. "If a horn player dies," says Dodge, "the orchestra stops playing."

Aiming for absolute reliability, Dodge and Bell decided to take a different route. Their OS would contain a "microkernel," a tightly written piece of software only a tiny fraction of the size of the big kernel. Instead of taking up megabytes of memory, the microkernel could be squeezed onto a single 40K chip. All the components of their OS were isolated from the microkernel and from one another in their own protected memory spaces. That makes QNX a "distributed system"--one in which the orchestra is not confined to a single space. And thanks to the program's "self-healing" feature, a dead player is automatically replaced or resurrected in millionths of a second without affecting the rest of the band. QNX has been the only company so far to commercialize a microkernel OS.

The microkernel enables QNX to run at blinding speeds. QNX also ranks functions, giving priority to the most critical, deadline-driven ones. Those features give it an unmatched ability to operate in "hard real time," responding to vast amounts of incoming signals in a few hundred nanoseconds (billionths of a second) and performing the right tasks at exactly the right moment. The capacity to "juggle ten billion balls a second without dropping one," as Bell puts it, is what makes QNX ideal for extreme manufacturing and other critical control operations. A QNX OS that runs materials-handling and assembly lines at the Siemens Dematic plant in Grand Rapids, for example, has been operating without failure for 13 years.

To pursue their dream, Dodge and Bell founded QNX in 1980 in a small office next to a shoe store in a shopping center in Kanata. They never looked for venture capital, and QNX is still privately held. Today the company has annual revenues of around $30 million, but at first the founders barely made it; Dodge was supported by his working wife, and Bell, then single, lived with his parents to save money.

Because of their industrial focus, Dodge and Bell managed to escape Microsoft's notice for most of their company's history. In recent years, though, Gates & Co. have made inroads into manufacturing markets; Microsoft has recruited outside software firms to adapt its NT operating system for use in less demanding manufacturing applications, such as running standard milling machines and lathes. And Microsoft's Windows CE is a sort of mini-operating system that can easily be adapted to run things such as machine tools. After that program was introduced in 1996, Dodge and Bell say, Microsoft executives told them they would run QNX out of business in two years.

Nothing of the sort has happened, obviously. Instead, new markets are opening up for the sort of dependability and versatility that QNX offers. And the company has found a powerful ally in IBM (not to mention other new partners, such as Intel, Motorola, and Toyota). Big Blue's Skip McGaughey, who has worked on making QNX the software behind IBM's new automotive computer systems, says the company chose QNX because it represents the "very best" of real-time operating system technology. "The typical automotive end user would have no patience with a unit that freezes up or experiences systems errors," he says. Wonder which archrival company's software he's thinking of.