Scientists tap DNA to try to build 'perfect computer'
Using chemicals instead of silicon, researchers take first step toward
A gold-plated piece of glass at the University of Wisconsin could hold the future of computers.
The device is rudimentary - it can only perform basic high-school-level math problems. But this tiny "biocomputer" no bigger than a dime is powered by DNA, not silicon, and it represents one of the first significant steps toward a holy grail of science.
From air-traffic patterns to human genetic data, DNA computers could theoretically have a nearly limitless ability to store and analyze certain types of information that befuddles even the most advanced silicon computers. They could, for instance, store the information contained on a trillion CDs in a gram of DNA.
Now, as researchers have begun to bump against the outer performance limits of silicon chips, these new advances promise to vastly expand the limits of what humans can compute and accomplish.
"The potential of DNA-based computation lies in the fact that DNA has a gigantic memory capacity and also in the fact that the biochemical operations dissipate so little energy," says University of Rochester computer scientist Mitsunori Ogihara. "The question is: Is it possible to take advantage of these two facts to be able to do something that one cannot do with silicon-based computers?"
The possibilities for DNA computers include developing a credit-card-size computer that could design a super-efficient global air-traffic-control system. Other researchers say the military could take blood samples from an entire army and encode it to form a database of each soldier's blood type and genetic data.
"You could place that in a very compact form and search that easily," says John Reif, a computer scientist at Duke University in Durham, N.C., who is overseeing a US Department of Defense research effort into DNA computing. "And that's just one of an enormous number of possible applications."
Traditionally, biology and computer science have remained separate entities. But at the most basic level, DNA seemed perfect for computation.
Whereas silicon computers use a binary number system that represents all calculations in strings of zeroes and ones, DNA contains information in strings of four nucleic acids, represented by the letters C, G, T, and A.
Silicon computers manipulate the information by changing the values of the strings of ones and zeroes. DNA manipulates the data through biochemical reactions that can alter or match up pairings of nucleic acids.
"In order to do computations, you only need two things: a means of storing information and a means of manipulating information," says Lloyd Smith, one of the scientists who worked on the project at the University of Wisconsin at Madison. "Any system that has those two properties one can set up to do those computations. And DNA has them."
The computer in Wisconsin, described today in the scientific journal Nature, is basically a forest of DNA tethered to the piece of gold-plated glass.
While this computer is man-made, scientists have long marveled at the way DNA stores and transmits information in the biological world.
"There exist many examples in nature of problems that are inherently computational. For example, when our immune systems search to find the specific anti-body to fight [an illness], that is an extremely complex computation," says Laura Landweber, a Princeton University evolutionary biologist.
The first breakthrough came in 1994 from University of Southern California professor Leonard Adelman. He first used DNA to solve the "traveling salesman" problem, a classic math treatise that considers how to devise the most efficient route for a traveling salesman. The salesman would have to leave from a specific city and end at a specific city, thus limiting the possible routes.
These types of problems can describe many crucial real world issues, such as how to break top-secret codes. But they have stymied traditional computers because each added variable - for example, each added city - exponentially increases the complexity of the problem.
Still, DNA computers will likely never replace desktop PCs. Instead, they will be used for specialized problems that have proved intractable to traditional silicon computers. For example, DNA computers could be made of organic materials and exist inside the body without harm to the system. Another possibility is to use them to solve complex combination problems like the salesman problem, albeit at much higher levels.
Scientists say that programming DNA computers could prove difficult because of the problems still inherent in manipulating DNA. Furthermore, the number of errors that naturally occur in DNA processes remains a large hurdle to creating reliable computing systems.
Then again, the field is only five years old. "If you think about the history of computing, real computers could not do that much after they were invented either," says Dr. Landweber. "There is quite a learning curve."
(c) Copyright 2000. The Christian Science Publishing Society