For instance, the researchers found that two different genes implicated in smell had been altered by DNA shuffling. In one individual, the two genes lay side by side; in the other, they had fused together to form a single gene. "The presumption is that this reflects differences in the way people smell things, although we don't know that yet for certain," says Snyder.

To reconstruct how these alterations formed, Snyder and his colleagues went back and sequenced the ends of the chunks of DNA that were either inserted or deleted. The team did this for some 200 of the variants identified. "That was quite impressive," says Stephen Scherer, a geneticist at the Hospital for Sick Children, in Toronto.

Many of the structural variants may have been the result of sequences of DNA that spontaneously jump from one place to another in the genome, explains Snyder. Other changes may have occurred through the unequal swapping of DNA between pairs of chromosomes.

Evan Eichler, a geneticist at the University of Washington, in Seattle, describes the work as a technological feat. "New DNA sequencing technologies are coming out fast and furious," he says. With 454's sequencing technology, he adds, the researchers obtained a large amount of information about the human genome in a relatively short period of time.

Snyder emphasizes that there are many more structural variants to be found, not only in their samples (the group's technique could not discern changes smaller than 3,000 bases), but also in the genomes of other individuals. Michael Egholm, 454's vice president of research and development, hopes to repeat the experiment on 100 individuals in a year, and then on 1,000. "Our technology is getting faster," he says. "We did this in a fraction of the time it took others to do this kind of analysis, and at higher resolution and at greater depths."

The ultimate goal, says Egholm, is to create a complete inventory of all the structural variants in the genome and then begin testing each one to see which are associated with disease.