Can We Stop the Next Killer Flu?

He considered studying the yellow fever that killed so many people in the 1800s. But then he seized upon the Spanish influenza of 1918. It was wildly infectious, and virtually everyone on the planet was exposed. About 2.5 percent of those who became sick died, which seems like a modest level of lethality until you realize that it added up to more than 600,000 American deaths in just a matter of months and something like 40 million deaths worldwide. Taubenberger knew that the institute had millions of autopsy specimens from soldiers dating to the Civil War. If he could retrieve even a few genetic scraps of that virus, perhaps he could figure out why it was so contagious and virulent.

Ten years later, his project is still going, centered in the rather ordinary laboratory directly next to his office (he has collaborators in labs around the country). Taubenberger doesn't do a lot of bench work these days, what with giving interviews, taking meetings, trying to get things published, but he has assistants busily at work, filling tiny vials with fluids containing DNA, sequencing genes, tapping on computers, accessing databanks and doing all the highly detailed work of decoding the 1918 influenza virus. Taubenberger also has a new project in collaboration with the National Institutes of Health and a nearby genomics institute, to find the genetic codes of many thousands of different strains of viruses harvested from people and wild birds.

The overarching goal for both projects is to learn how these viruses evolve and which mutations might make them more or less likely to become adapted to humans and develop into potential killers. By removing from influenza some of its element of surprise, we might be able to forecast likely outbreaks, in the same way that we can forecast which tropical depression is going to turn into a hurricane. It's a sweeping plan, using all the hardware Taubenberger can round up.

If you take a left out of his lab, go through another lab (more vials, bottles, jars, tubes, refrigerators) and cross another hallway, you'll reach the room with the automated sequencers. There's a big one from Applied Biosystems, the 3132 Genetic Analyzer. Somehow, this thing can read the language of a genome, letter by letter.

Life on Earth operates on a genetic system that, at its core, is remarkably simple, considering that it gives rise to creatures as diverse as sea urchins, praying mantises and humans. The genome is written out on a very, very long molecule called deoxyribonucleic acid -- DNA.

Molecular biology is to some extent the study of architecture. It's all about structure. Proteins -- which do most of the heavy lifting in the body, such as building cells and tissues -- have many ways of folding themselves in three dimensions. Their structure determines their function. They roam the body in search of a correctly shaped receptor. They just want to fit in somewhere.

When Francis Crick and James Watson rocked the scientific world in 1953, it wasn't by discovering DNA. Rather, they found the structure of the molecule, and proved that it was the source of genetic information. "We've found the secret of life," Crick exulted that winter day to friends at the Eagle pub in Cambridge, England, and the secret, it turned out, wasn't some special juice, some exotic energy source, but just a well-framed, two-stranded, ladderlike molecule with rungs in all the right places and a nifty ability to make copies of itself.

A gene is historically defined as a segment of DNA with instructions for making a single protein, though the one-gene, one-protein rule is pretty loose. Humans have upwards of 30,000 genes. The flu virus has just 11.

The code of a gene is written in the form of tiny chemicals called nucleotides, more commonly referred to as the "bases" or "letters" of the genome. Life uses a very short alphabet. There are only four bases used by living things: adenine, cytosine, guanine and thymine, or A, C, G and T.

DNA sequencing, the process of finding the order of the letters, isn't terribly new. As far back as 1977, Fred Sanger and colleagues managed to piece together all 5,386 letters of a tiny organism called phi-X174. In the mid-1980s, Kary Mullis developed a technique still used in Taubenberger's lab, called polymerase chain reaction, which amplifies pieces of DNA and makes them easier to study.

Automated sequencing machines came online only in the past decade or so. They're like reverse vending machines. You open a door, place a tray of DNA samples in a slot, watch it recede into the interior of the machine, and wait. Inside the machine, needles descend into the DNA vials and pull the fluid through a tiny glass tube, known as a fiber-optic capillary. The machine examines the thin stream of fluid with laser light; the nucleotides, the bases, go slipping through the laser beam one by one, guanines glowing differently from cytosines, and so on. Soon, the results flash on an adjacent computer screen: the letters. The code. The process is hardly push-button simple -- the machines can examine only short segments of genes at one time, and scientists are often working with scraps to begin with. But it's definitely a scientific marvel.

"There's this kind of voodoo part," Taubenberger says. "Nothing you do can be seen. It's all invisible. It's all magic." But, he adds, in homage to the requirements of the scientific method, "it's reproducible magic."