THE calculating power of billions of molecules has been tapped for the first time in a revolutionary new type of 鈥淒NA computer鈥�. Leonard Adleman of the University of Southern California encoded the 鈥渋nput鈥� details of a computationai puzzle into single DNA strands, which he then mixed together, allowing them to link up into many different double helix 鈥渙utput鈥� molecules among which a few had a structure encoding the answer (Science, vol 266, p 1021).
The puzzle involved finding a specific path through a network of points. An example might be as follows: take the four cities London, Paris, Berlin and Amsterdam. Non-stop flights are scheduled only from London to Paris, Paris to Berlin, Paris to Amsterdam and Berlin to Amsterdam. By which route from London to Amsterdam could a traveller visit all cities, taking only three journeys?
In this case the answer is obvious: fly from London to Paris, then to Berlin, then on to Amsterdam. But if the problem involved all the major cities in the world, and all the connecting flights, the number of possible itineraries would become astronomical. On a large scale, this type of problem is impossible to solve even with the fastest supercomputers, as the time taken to find the result grows exponentially with the number of destinations.
Advertisement
Adleman solved a seven-city problem by encoding the details into single strands of DNA. The 鈥渄ouble helix鈥� of DNA is formed of two complementary strands of the four DNA bases, labelled A, T, G and C. The base A is the complement of T 鈥� that is, it can bind only to T. Similarly, G is the complement of C. One strand is therefore a complementary mirror of the other. Adleman randomly selected single strand codes to represent each city 鈥� say ATGCGA for London, TGATCC for Paris, GCTTAG for Berlin. For example, the strand representing each flight path might be defined by the last three code letters of the city the flight was leaving, and the first three code letters of the destination. So London to Paris would be coded CGATGA in the four-city example.
Using genetic engineering, it is possible to manufacture single DNA strands to order. Adleman mixed the complement strands of the city DNA (here ACTAGG for Paris, say) with the flight path strands in the test tube, and as they joined to form double helices, the flight path strings acted as complementary 鈥渟plints鈥� to bind the DNA city strings together. Molecules for all possible combinations of flights formed, but given that billions of molecules were reacting, it was almost certain that a molecule representing the correct flight path would be in the final mixture.
The problem was to identify the 鈥渙utput鈥� molecule. It has the property that it spans the length of all the city codes, beginning with the starting city and ending with the destination, and it contains the codes of all other cities en route only once. Knowing this, Adleman was able to isolate the molecule representing the solution using standard techniques of molecular biology. He then analysed the order in which its building blocks had been put together, revealing the correct sequence of cities.
One exciting aspect of the 鈥淒NA computer鈥� is its potentially unprecedented energy efficiency and minimal 鈥渕emory鈥� requirements. 鈥淪toring information in DNA requires about one trillionth the space required by existing storage media such as video tape,鈥� says Adleman. However, he points out that in its present form, this method of computing is not practical enough to replace conventional computers; the solution to his seven-city problem emerged only after seven days of work. But if future research can find short-cuts, biological computers may become a practical method for solving complex 鈥渘etwork鈥� calculations often encountered in real life, such as the design of telephone networks.
Adleman also hopes to explore some of the more philosophical spin-offs of molecular mathematics. The cells in the body are host to incessant DNA and RNA reactions, which viewed as 鈥渃alculations鈥� might seem utterly meaningless 鈥� or are they? Nature鈥檚 鈥渃omputing鈥� may play a role in the mechanisms which underlie biological processes such as evolution and the functioning of the immune system.