A molecular computer that plays tic-tac-toe

I remember, when I was a (probably irritatingly nerdy) child, being absolutely fascinated by making a tic-tac-toe playing automaton out of match-boxes and beads, following a plan in one of Martin Gardner’s books. So my eye was caught by an item on Martyn Amos’s blog, reporting on a recent paper in Nano Letters (abstract and graphic freely available, subscription required for article) from a group in Columbia University, demonstrating a tic-tac-toe playing computer made, not from matchboxes or even more high-tech transistors, but from individual molecules.

The basic logic gate of this molecular computer is a single short DNA strand of a prescribed sequence which can act as a catalyst – a deoxyribozyme. Like the protein molecules used in the molecular computing and signalling operations inside living cells, these molecular logic gates operate by allostery. This is the principle that when one molecule binds to the gate molecule, it changes its shape and makes it either easier or harder for a second, different, molecule to bind. In this way you can get differential catalytic activity – that is, you can get a situation where the logic gate molecule will only catalyse a reaction to produce an output if a given input molecule is present. This simple situation would define a gate that implemented the logical operation YES; if you needed two inputs to stimulate the catalytic activity, you would have an AND gate, and if you have an AND gate whose catalytic activity can be suppressed by the presence of a third molecule, you have the logical operation xANDyANDNOTz. It is these three logical operations that are integrated in their molecular computer, which can play a complete game of tic-tac-toe (or naughts and crosses, as we call it round here) against a human opponent.

The Columbia group have integrated a total of 128 logic gates, plausibly describing it as the first “medium-scale integrated molecular circuit”. In their implementation, the gates were in solution, in macroscopic quantities, in a multi-well plate, and the outputs were determined by detecting the fluorescence of the output molecules. But there’s no reason in principle at all why this kind of molecular computer cannot be scaled down to the level of single or a few molecules, paving the way, as the authors state at the end of their paper, ” for the next generation of fully autonomous molecular devices”.

The work was done by Joanne Macdonald and Milan Stojanovic, of Columbia University, and Benjamin Andrews and Darko Stefanovic of the University of New Mexico – there’s a useful website for the collaboration here. Also on the author list are five NYC high school students, Yang Li, Marko Sutovic, Harvey Lederman, Kiran Pendri, and Wanhong Lu, who must have got a great introduction to the excitement of research by their involvement in this project.