Complex Calculation No Problem for DNA ‘Computer’ : Mathematics: Mixing a unique molecule and the proper enzymes in a test tube, a researcher uses the natural chemistry in one of life’s building blocks to solve an equation.

Researchers have found a way to use the language of life itself--the spiral string of DNA curled inside every living cell--to solve a difficult math problem, suggesting that one day a working computer could be crafted in a test tube.

USC computer expert Leonard M. Adleman used the genetic code to program an elementary equation into a unique DNA molecule. By combining the new molecule and the proper enzymes in a test tube, he used the natural chemistry underlying all living things to arrive at the correct solution.

In a sense, Adleman evolved the right answer.

For the moment, the provocative experiment is a laboratory novelty. But David K. Gifford, a computer scientist at the Massachusetts Institute of Technology, suggested that once the tools of the molecular machine shop have been refined, it may “revolutionize the way we think about both computer science and molecular biology.”

Theoretical biologist Stuart A. Kauffman, an expert in genetics and complexity at the Santa Fe Institute, called the experiment “fascinating.”

The research, made public today in the journal Science, involves a wide-ranging class of computations encompassing hundreds of normally intractable problems that crop up in artificial intelligence systems, complex computer networks such as the budding information superhighway, or intricate games such as chess. Such problems, which involve selecting the correct path through an enormous number of possible choices, usually are more than the fastest supercomputer can digest.

To program his molecular computer, Adleman encoded the problem by arranging the chemical units of an artificial gene sequence, just as a more conventional programmer orchestrates the digital bits in an electronic computer register. To carry out the calculation, he took advantage of DNA’s natural ability to rapidly reorganize itself and recombine in complex but predictable patterns.

Through the millions of random combinations and recombinations in the test tube, a complex DNA molecule emerged that contained the correct solution encoded in its genetic sequence. The chemical reaction produces billions of variations, essentially trying all possible solutions by brute force.

To recognize, isolate and then multiply the right molecule, Adleman used the tools of modern molecular biology in essentially the same way that police identify the DNA in blood stains at a crime scene. With a series of special enzymes, Adleman screened for the sequence that encoded the answer to the calculation. Once it was isolated, he made millions of copies of it through the polymerase chain reaction (PCR) technology, which is used in many criminal cases.

“The main idea of this is to have all those millions of little molecules all in parallel, running around trying out solutions to your problem,” he said. “Because there are so many, the vast numbers make the odds more favorable that you will accidentally come across the solution.

“You use the parallelism inherent in chemical reactions,” he said.

Although the computation itself took no more than a second or two, Adleman said, it took seven days to extract the answer from his test tube of DNA.

A biological computer based on Adleman’s technique may never be as flexible as the electronic PCs that have multiplied across millions of desktops in the past decade, researchers said. In principle, however, a calculating device that relied on biology instead of electrical engineering to manipulate complex calculations would be a billion times more energy efficient and could store information in one trillionth the space.

Gifford cautioned that the experiment is no more than a first step on the path toward computation with DNA. So many obstacles remain that the technique is not practical enough to replace a more traditional computer even for the calculation Adleman has already solved.

However, “once these limitations are removed, biological methods could become a practical method for solving real-world computational problems,” Gifford said. “The simplicity of Adleman’s method is surprising given that it solves a hard computational problem.”