In a traditional computer, data are represented by and stored as strings of zeros and ones. With a DNA computer, a sequence of its four basic nucleotides — adenine, cytosine, guanine, and thymine — is used to represent and store data on a strand of DNA.
Calculations in a traditional computer are performed by moving data into a processing unit where binary operations are performed. Essentially, the operations turn miniaturized circuits off or on corresponding to the zeros and ones that represent the string of data.
In contrast, a DNA computer uses the recombinative properties of DNA to perform operations. The best way to understand the processing power of recombination is to look at an example.
One of the first problems tackled back in 1994 by Leonard Adleman and his team at the University of Southern California used DNA computing techniques to find a route between seven cities. This is a classic math problem belonging to a family known formally as the Hamilton path problem, often referred to as the traveling salesman problem.
The gist of the problem: Given a starting point and an end point, find the shortest route through all seven cities without hitting any city more than once.
In the DNA computing approach to this problem, each city was represented by a unique strand of DNA that stretched 20 nucleotides long. Each possible route between any two cities was represented by another 20-nucleotide-long strand. This strand connecting cities was also related to the nucleotide sequence of each city connected by that route. For example, the route between City 1 and City 2 consisted of two sets of 10 nucleotides; the first 10 nucleotides on the strand's route complemented the last 10 nucleotides of City 1, while the second 10 nucleotides on the strand's route complemented the first 10 nucleotides of City 2.
In this way, if strands for cities 1 and 2 came in proximity to the route 1-to-2 strand, the three strands would bind.
To solve the traveling salesman problem, strands representing all seven cities and strands representing all possible routes between any two cities were placed in a test tube. The end result was a series of longer, recombined strands.
The correct answer was contained in a strand that started with City 1, ended with City 7, contained the strands of all seven cities, and had no one city represented more than one time in this longer strand.
When the experiment was first done in 1994, the complexity of finding the strand with the correct answer was considered a downside to DNA computing.
"In my experiment, one molecule had the right answer, but several trillion molecules had the wrong answer," Adleman said in a 1996 interview. According to Adleman, it took a week of work to extract the right answer. But he also noted that he was doing all the work by hand.
Back to Calculating with DNA