The wet and wild future of computers [DNA strands may replace chips and electrons ]greenspun.com : LUSENET : Human-Machine Assimilation : One Thread
The wet and wild future of computers DNA strands may replace chips and electrons
John S. MacNeil U.S. News & World Report 02/14/2000
The element silicon is so closely identified with computers that most people would be likely to associate it more readily with California's high-tech valley than with the periodic table. But such thinking may soon have to be radically revised, as high-speed computation moves beyond chips and machines to include the tools of biochemistry and genetics: test tubes, slides, solutions--even DNA.
DNA is present in every living organism, and the appeal of the molecule as a supercomputer mechanism lies in its demonstrated ability to store a vast amount of information--indeed, all of the instructions for replicating life. Although the chemistry set won't be replacing your PC anytime soon, two groups of scientists demonstrated last month how these information-laden molecules might perform calculations in future computers.
Instead of using zeroes and ones to encode information using electrical current, the "memory" in a DNA computer takes the form of thousands of DNA strands that are synthesized in a lab. Each strand contains a different sequence of the chemical bases--symbolized by the letters A, C, T, and G--that make up all DNA molecules. To sift through all these strands, scientists subject the DNA memory to various enzymes that eliminate certain strands of DNA, leaving only the strands of bases that represent correct answers (graphic).
In January, scientists at the University of Wisconsin reported in the journal Nature that they had found a way to perform a simple calculation using strands of DNA that had been attached to a gold- plated surface. Previous experiments with DNA computing had allowed the DNA to float freely in a test tube, but Lloyd Smith, a chemist and leader of the Wisconsin research team, hopes his method will allow the wet chemical steps required for a calculation to be automated. "It's a route to scaling up DNA computing to larger problems," says Smith of his experiment.
Chess and chemistry. Another group, led by biologist Laura Landweber at Princeton University, reported on a way to use RNA--a chemical cousin of DNA--to perform a similar calculation. To demonstrate that their technique works, Landweber's team calculated the answer to a simple version of a classical chess dilemma called the "knight problem." The computer must determine in which positions a chess player can place the knights on the board so that none can attack another. The scientists encoded each strand of RNA to represent a possible configuration of knights. Then, they performed a series of steps in a test tube with chemicals designed to eliminate RNA strands representing wrong answers, and then they analyzed the remaining RNA strands to see if they all corresponded to correct answers. Almost 98 percent of the supposedly "correct" strands did in fact correspond to correct chess configurations--a surprisingly high success rate for a preliminary experiment.
At the moment, it is still much faster to use a PC to perform such calculations. But silicon-based computers perform their magic simply by running through every possible answer one by one at the speed of electrical current. Because of DNA's power to store information--a few grams of the material could store all the data known to exist in the world--some scientists believe that such biochemicals will eventually be the most efficient medium of storing and manipulating information. But its real advantage over a conventional computer is that rather than analyzing each possible answer in sequence, the DNA computer would act on the entire library of molecules--or answers-- simultaneously.
Although Landweber is optimistic about the ability of the technique to find the right answers with 100 percent accuracy in the future, she and other researchers are quick to point out that the field is in its infancy compared with conventional computing methods and that for many applications, silicon-based microchips will always be better. "Silicon computing won't go away, and the applications that it's used for won't go away," says John Reif, a computer scientist at Duke University and director of the Consortium of Biomolecular Computing.
What's really needed, according to most researchers in the field, is a "killer" application particularly suited for the way DNA computing solves problems. Such real-world problems might involve the encryption of large amounts of military information, or they might involve some combination of silicon and DNA computing, says Reif. "We'll just have to see how far we can push the technology--see how far we can take it," says Landweber.
The biochemical computer
Scientists are making "computers" out of strand of synthetic DNA rather than silicon chips. These computers perform computations simultaneously rather than one by one, as today's machines do, making them potentially much faster. Here is one method from the University of Wisconsin:
Sources: Lloyd Smith and Robert Corn, University of Wisconsin
-- scott (firstname.lastname@example.org), February 15, 2000