by David Bradley

Prasanna 'AP' de Silva and Nathan McClenaghan at Queen's University, Belfast have used molecules that switch on and off depending on a chemical signal, and can produce a tiny burst of light representing a binary number 1. Such molecules can work as the equivalent of the tiny logic gates on a silicon chip, making decisions on each bit of computer information that passes through them. By running them together the researchers can carry out arithmetic operations.

Normally, a vast array of millions of transistors etched on to a silicon chip is at the heart of a computer, but chips are like whole cities on the molecular scale and when real estate is getting tighter and tighter it is time for the molecular architects to step in.

'The extension of information processing and computation to the molecular level will only be possible when molecular logic gates are available,' according to Vincenzo Balzani of Bologna University. He believes chemists can design and construct molecules to do the work of silicon chips but in a much smaller space and far faster.

There are several different logic gates used in computing. For instance, an 'AND' gate gives an 'on' or '1' output when both its inputs are '1'. An 'OR' gate, is only 'on' or '1' when its inputs are '1' or '0'. The slightly bizarre 'XOR' (eXclusive OR) gate acts like a 'spot the difference' unit giving no output if both inputs are 1s or 0s but a '1' if they are '1 and 0' or '0 and 1'. The NOT, OR, INH, and NOR gates and many others all work together to do addition, subtraction, multiplication, and division. With those mathematical functions, you can carry out any conceivable computer process - from displaying a single word on-screen to rendering a multimedia web site. Computation always reduces to maths.

De Silva and his colleagues previously built molecular versions of several logic gates, including the AND and XOR. The gates are based on tiny claw- shaped molecules that trap different ions. They then use these ions to represent binary 1's. When two different ions are present (a calcium and a hydrogen) the claw molecule glows blue, representing a '1' output. With no ions present the molecule stays switched off, there is no glow, so the output is 0. The XOR molecule works similarly but lights up only when the two inputs are different.

The clever bit, explains de Silva, comes when you makes these two logic gates work together. Say you're adding two numbers (in binary) - 1 and 1. When the ions representing these numbers are present and are trapped by the AND gate it lights up - the output is a 1. You can call this the 'carry over' number of primary school sums. The XOR with the same combination of ions does not illuminate it gives a 0. So, 1 add 1 equals zero, carry one, which in binary is '1-0'. In other words 1+1=2.

PhotosToGo DNA is one of the most efficient digital systems known

'Such logic operations are the basis of all computing,' explains de Silva, ' by using molecules instead of silicon chips we hope to be able to perform small-scale computational operations in very small spaces'. Scientists are on the verge of being able to manipulate and observe single molecules using fluorescent sensors, and will soon be able to 'see' the glow from individual logic gates. De Silva adds that, 'The first "real" device applications are expected to occur fastest in the fields of biotechnology and combinatorial chemistry where small volumes are common.'

Nature, though, may have provided scientists with an alternative to building logic gates, at least for some kinds of 'number crunching' computational problems.

The genetic code formed from the nucleic acids, RNA and DNA, is one of the neatest digital information systems we know. In the early 1990s, Leonard Adleman at the University of Southern California began figuring out how this digital code might be used to solve mathematical puzzles, such as the travelling-salesman problem. Adleman used different strands of DNA to represent different routes a salesman might take between a group of towns so that his total journey time is kept to a minimum.

A set of enzymes that can single out those sequences corresponding to a shorter journey were then used to split, or cleave, the DNA strands that are not optimum, step by step. Eventually, the array of combinations is reduced to a single strand representing the best route.

PhotosToGo How many ways can two opposing knights be placed on a chessboard so they are not attacking each other? Could the answer be found with an RNA calculator?

Earlier this year, two other research groups have applied similar thinking to a second puzzle that asks the question, 'How many ways can two opposing knights be placed on a chessboard so they are not attacking each other?'

Laura Landweber collaborated with Richard Lipton, Dirk Faulhammer and Anthony Cukras to see whether they could solve this chess problem using DNA's chemical cousin ribonucleic acid (RNA). To make the experiment simpler they reduced the chessboard to a 3x3 grid and used different strands of RNA to form combinations each representing a location of the two knights on the board.

An enzyme - ribonuclease - was then used to seek out specific combinations of bases along the strands of RNA and cleaving those that do not match the correct answers to the problem. They came up with most of the answers in a very short time but the fact that it works alone demonstrates the potential of an RNA calculator.

Meanwhile, Lloyd Smith and his colleagues at the University of Wisconsin- Madison demonstrated a slightly different approach to the same puzzle using DNA itself. However, their system is built on a solid support, which lends itself better than dissolved molecules, to more direct connection to external devices, such as electronic components.

Of course Science Fiction writers got there first, the computer system of the latest Star Trek ship (Voyager) is biological - only a molecular system is small and powerful enough to run a ship of that complexity!

With logic gates adding up and strands of DNA solving age-old maths puzzles, it is perhaps only a few more years before chemistry will come up with a prescription for a molecular computer.Thorri Gunnlaugsson a former member of de Silva's research team and now heading his own group at Trinity College Dublin asks us to imagine a memory chip the size of a sugar cube carrying as much information as a thousand billion CD-ROMs or a molecular chip running a thousand times faster than the PC on your desktop. Sadly, there might then come a time when a computer crash doesn't just leave you frustrated by a blank screen but might soak your desktop too!