About 100 million years ago, a creature the size of an opossum ambled through the forests of what is now South America. It was probably a ratlike thing, with coarse fur, a scrawny tail, and furtive eyes. If you went back in time with a .22, you could pick it off with one well-aimed shot. But that wouldn't be a good idea. That creature was your ancestor.

 

Over millions of years, an evolutionary cornucopia spilled from that unassuming ur-mammal. The species to which it belonged split into two daughter species, and then those species split, and the process repeated again and again. One line eventually led to rabbits, beavers, and mice. The members of another line began hunting in shallow bodies of water and gradually evolved into whales and dolphins. Meanwhile, with a few exceptions, the other mammals living back then - and their descendants - eventually went extinct.

In his office overlooking the redwood groves of UC Santa Cruz, David Haussler eagerly shows me our pedigree. "Here's the common ancestor," he says, writing the word Boreoeutherian at the top of a piece of paper. He draws downward-­branching lines with animals at the tips. "Here's us," he says, filling in the last two labels - chimpanzee, human.

Biologists have been drawing diagrams like these since Charles Darwin sketched the first evolutionary tree in 1837. But Haussler's reconstruction process is ­different. Instead of examining fossils and tracing a line from extinct creatures down to those alive today, he's trying to move back up the evolutionary tree. Haussler is attempting to run evolution in reverse.

He starts by comparing the genomes of humans and other existing animals with one another, making inferences about the DNA sequences in their common ancestors. Haussler has used this technique to mathematically reassemble parts of the genome of the progenitor of chimps and humans - a shambling, hirsute, apelike creature that lived about 6 million years ago. He has reconstructed DNA sequences of the predecessor of most hoofed animals, an unprepossessing beast that had to dodge the footfalls of dinosaurs to survive. Most audaciously, Haussler and his collaborators have pieced together much of the genome of the ur-mammal itself, which they plan to release in draft form later this year. "Haussler can reconstruct its genome with a fairly high accuracy" says Eric Lander, director of the Broad Institute and a leader of the public Human Genome Project, "and that's way cool."

Haussler's unexpected success complements a frenzy of work done by researchers using other methods to determine the genetic makeup of extinct organisms. Last year, scien­tists working with physical DNA specimens published the sequence of a big chunk of a genetic code extracted from a frozen woolly mammoth bone. Another team recovered 40,000-year-old DNA fragments from cave bears. Other groups have gone after the DNA of extinct plants, insects, and even dinosaurs.

Wait a minute. Wasn't all this "ancient DNA" talk pretty well trashed after Jurassic Park? When an animal dies, DNA starts to break down like a cigar left in the rain, and, after the movie came out, ­scientists showed that amber-encased mosquitoes would never be able to provide enough dinosaur DNA to re-create a T. rex.

But the past few years have brought new developments. Scientists have gotten better at isolating DNA from fossils. They have also learned that perfectly preserved samples aren't necessary to build up lost genomes. Meanwhile, Haussler, benefiting from clever algorithms and massive increases in computing power, has made it much easier for them to fill in the gaps. If one scientist has sequenced DNA fragments from a woolly ­mammoth bone, and if Haussler has a tool that can re-create other parts of its genome, the two together put us a lot closer to seeing that beast at the local zoo.

Haussler insists that he just wants to explore human evolution and solve medical mysteries. "The goal is to understand life, not to create a Jurassic Park," he says. But put the genome of an extinct organism in a computer database and it will cry out to be rebuilt. Doing so could produce valuable insights into evolution - like why humans are susceptible to some diseases that other primates aren't - and many biologists think it's an experiment we're getting closer to being able to run. Hendrik Poinar of Canada's McMaster University and his father, George, an expert on amber-preserved biological samples, were consultants to Steven Spielberg on Jurassic Park. "People kept asking us, 'Is this ever going to happen?' and we would say, 'No, it's never going to happen,'" Poinar recalls. "But the picture is somewhat different now."

If there's a member of our extended family that Haussler resembles, it's the camel. He's tall, blond, and broad-shouldered, with a ruddy complexion. A self-described math nerd, he looks like a surf bum who has spent too much time in front of a computer screen.

Haussler grew up in the San Fernando Valley outside of Los Angeles. A propeller­head as a kid, he became disenchanted with ­science and mathematics in high school and enrolled in tiny Immaculate Heart College in Hollywood, thinking he might become an artist or musician. But then he took calculus and rediscovered astronomy. "I thought, 'Wait a minute. Why did I turn my back on this?'"

By the early 1980s, he was working on his computer science PhD at the University of Colorado at Boulder. One day, another grad student told him about biologists who had just started sequencing long stretches of bacterial DNA and needed a way to locate particular patterns. That intrigued Haussler, whose specialty was writing algorithms to sort through data. He began working with DNA sequences and eventually became a full-time biologist.

In 1999, he joined the public Human Genome Project. And that's when the reverse-evolution machine began taking shape. As the project was winding down, Haussler and several other programmers working in the same lab built a browser that made the genome available to anyone - essentially open-sourcing their data. The browser quickly evolved. Once the human genome was complete, scientists put their sequencers to work on the genomes of mice, rats, dogs, chimps, and other organisms. Some sections were similar, reflecting their descent from a common ancestor; others were different, indicating the effects of evolution.

That got Haussler thinking. Scientists had reconstructed the sequences of individual genes from extinct species. But no one had even started working on re-creating an entire genome. Of course, the genomes wouldn't always line up - evolution re­arranges them over time. But the fragments could still be compared. And evolution tends to preserve exactly those parts that are most important.

Here's an analogy: You ask 10 friends to remember the letter G. But the next day you discover that some, including you, have forgotten it. When you ask all 10 what the letter was, four say "G," while the others choose random letters. Since "G" is the most common response, you can pretty safely assume that G is the letter you told them. Do the same thing several billion times with the DNA sequences of mammals that exist today and you should be able to determine the genome of the common ancestor from which those mammals evolved. The more genomes you feed into the model, the more accurate your result will be.

One of Haussler's graduate students, Mathieu Blanchette, tested out the technique. Using a sequence of virtual DNA as complex as a real genome, he programmed his computer to make the sequence evolve in a way that mimicked nature. He then used the "descendants" to try to reconstruct the original genome. The results astonished Blanchette, who is now a professor at Montreal's McGill University. "It actually worked."

Haussler, Blanchette, and their ­collaborator, Webb Miller at Penn State, hope to release the program they've developed into the public domain later this year, allowing anyone to build the genomes of extinct animals. Haussler expects the reverse-evolution machine to "keep people busy for a long time."

Biologists can give you lots of reasons why ur-mammals won't roam the earth again anytime soon. For starters, genomes are really long. A typical mammalian genome contains billions of base pairs. Geneticists have no idea, at present, how to construct DNA sequences of such length and insert them into cells.

There's another big issue: mistakes. Haussler estimates that he could determine the ur-mammal genome with 98 percent accuracy. But of course there's no way to double-check without the original DNA. Plus, 2 percent is a lot. A human genome that was 98 percent correct would still contain 120 million errors, any of which could cause horrific problems.

The genomes of some extinct animals will be much harder to reconstruct than others. The ur-mammal has lots of present-day descendants, which is why Haussler chose it as his initial target, but dinosaurs don't. Reconstructing the genome of a Tyrannosaurus rex would therefore require inspired guesswork based on the genomes of related species like birds and turtles, as well as DNA fragments recovered from fossils. (And suddenly we're back in Jurassic Park.)

Then there are the unanticipated problems that come about when you fool around with nature. "There could be unforeseen interactions between an extinct species we bring back to life and ourselves," says Christos Ouzounis, an expert in computational genomics at the European Bioinformatics Institute in Cambridge, England. And even if we could re-create, say, a bronto­saurus, it would be plunked down in a place where it didn't belong and where it would have no adults to teach it how to be a proper brontosaurus.

Are any of these objections show­stoppers? Probably not. Biologists have already succeeded in reconstructing viruses - organisms so simple that whether they're alive is a matter of semantics. The next, much harder step will be to build microorganisms. While biologists need to know a lot more about how cells work to do that, they can already modify an existing microbe or virus to create an earlier version of that organism - scientists recently rebuilt a strain of the 1918 flu that killed more than 50 million people.

Resurrecting extinct species will be much more difficult, but the prospect now exists. Researchers continue to get better at extracting DNA from ­fossils, and Haussler's reverse-engineering technique will become commonplace as more genomes from modern-day organisms are sequenced. According to Miller, within the next couple of centuries humans should be able to make any creature they want.

For now, Haussler and his colleagues are focused on more immediate, though still ambitious, goals. They plan to explore the functions of ancient DNA segments by bioengineering them into mice, and they'd like to identify the specific genetic changes that transformed the ur-­mammal into an upright, hairless, big-brained primate. But in the long run, Haussler says, the potential is unlimited. "These are scientific opportunities that rarely come along in a person's lifetime."