������ý

By making DNA endlessly change, researchers have shown how a biological computer might one day solve problems much faster than conventional computers or even quantum computers. It’s still a long way from being functional though.

The DNA-based system is an experiment in how it may be possible to make a theoretical type of computer known as a non-deterministic universal Turing machine.

Such a machine could solve tricky problems much faster than existing computers. Imagine that a computer is trying to find the centre of a maze and has a choice between left and right. A conventional computer would turn in one direction and follow that path to the end, then try a different route if that one leads nowhere.

But a non-deterministic universal Turing machine would explore both paths simultaneously, and do so again every time the path splits until it has found the right route to the maze’s heart.

Because these theoretical machines explore all possible routes to a solution simultaneously, they aren’t constrained by time in the same way that conventional computers are. With conventional machines, the trickier a problem is to solve, the more steps it requires, and the longer it takes to come up with an answer.

But, for non-deterministic universal Turing machines, trickier problems would take the same time to solve as easier ones. Their computational power is instead limited more by the amount of space they take up. Conventional processors pack billions of tiny transistors on a chip, but that’s nothing compared with the possible processing power of DNA.

Endless rearrangements

By using DNA, “it’s plausible to have on your desktop more processors working than exist in the rest of the planet put together”, says , who studies machine intelligence at the University of Manchester University, UK, and who led the team that created the system exploring how DNA could form the basis of a non-deterministic universal Turing machine.

King and his colleagues based their concept on the premise that a single strand of DNA can be endlessly rearranged using gene-editing techniques. By mixing DNA in a test tube containing seven different gene-editing molecules, they could effectively ensure that the genetic code in the DNA would be reordered into a huge number of new random combinations.

You could see the simultaneous creation of a vast number of different code combinations as the molecular equivalent of the computer exploring all routes in the maze at the same time. But there’s a big problem: you can’t program it.

This means the team’s system isn’t capable of tackling any problems, and King acknowledges that it’s a long way away from a working non-deterministic universal Turing machine. However, he says that his experiment shows it might be possible to use DNA to build such a computer.

Reality check

at the European Bioinformatics Institute in Hinxton, UK, is less convinced. “DNA is really hard to control,” he says, and might not be an ideal substance to use for the future of computing.

And at the University of Oxford says the jump between a DNA computer that can theoretically solve problems and one that can do it in practice is huge. A more realistic application of DNA computing, she says, might be DNA-based logic gates. These use short strands of DNA to act as tiny biological computers within cells that trigger the release of a molecule when they detect certain biomarkers within the cell.

It doesn’t look like DNA computers will be replacing silicon chips any time soon.