- guardian.co.uk, Thursday July 24 2008 00.01 BST
- The Guardian, Thursday July 24 2008
A scientist uses DNA to identify remains. Photograph: Getty Images
No, it won't happen. Even the people who are working on ways to make DNA perform calculations cannot see it happening. Take Professor Andrew Ellington at the University of Texas at Austin, who said: "I think my general opinion about DNA computing is not 'no' but 'Hell, no'."
So, how come some people believe that the future of computing lies in the molecule that provides the blueprint for how living creatures are made?
It started in the 1990s when Leonard Adleman, of the University of Southern California, came up with a way to use DNA to solve one of computing's most difficult and complex tasks: the travelling salesman problem.
This tries to find the most efficient way to visit a set of cities so that you pass through each one as few times as possible - ideally, only once. Adleman worked out that the paths between cities could be coded into DNA.
If a path between two cities matched, those DNA strands would stick to each other. Eventually, lots of strands would match up to produce a list of cities in DNA, packed into a classic double helix. Just shaking up a test tube full of those different DNA strands did the job. It was the ultimate parallel processor.
The idea of the DNA computer got another boost in the last month when Japanese scientists succeeded in making a type of DNA out of molecules not found in nature. People pointed to the fact that you could pack thousands of exabytes into the volume of a drop of water with regular DNA. Artificial DNA could, potentially, store even more.
Making all that DNA is another matter, however. In a silicon-based computer, the transistors that do the calculations are re-usable. You just load up another program and feed it to them. With a DNA computer, you have to make a complete new set of DNA strands each time. It's a new computer every time you want to calculate something. And making artificial DNA is not at all cheap.
It gets worse. DNA computing only looks fast because things can happen in parallel. But the reactions themselves are very slow, and the process only works if you can tolerate errors, said Ellington. Not only that, it's hard to work out what the answer was: there is no easy way to wire up DNA to the other parts of a regular computer.
That does not mean there are no ways to use DNA in calculations. Instead of trying to recreate computers, researchers aim to use DNA in simple biological machines: in principle, they could use DNA logic to administer drugs only to specific cells.
"DNA computers can manipulate matter: chemistry is their world," said Zack Booth Simpson, a software engineer and molecular biology researcher at the University of Texas. But when it comes to manipulating digits, silicon is likely to keep its edge into the foreseeable future.
