On supporting science journalism
If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
It's not your normal, electronic silicon-based machine, but scientists have made a computer from a small, circular piece of DNA, then inserted it into a living bacterial cell and unleashed the microbe to solve a mathematical sorting problem.
"A computer is any system that can read some input and give some readable output," says Karmella Haynes, a biologist at Davidson College in North Carolina and co-author of a new study appearing in the Journal of Biological Engineering. Haynes and her team looked to harness the power of DNA recombination to solve the so-called "burnt pancake problem": a puzzle about how to stack different-size flapjacks that are burned on one side and perfectly cooked on the other using the fewest number of flips to arrange them so the largest are on the bottom and all are golden side up.
"This work is the first work I've encountered which uses living cells in order to solve a specific computer science problem," says Tom Ran, a graduate student in the lab of computer scientist Ehud Shapiro at the Weizmann Institute in Rehovot, Israel.
By showing that DNA functioning as a computer would be able to solve the burnt pancake problem, Haynes and her team demonstrated that if their system could be scaled-up, it could spit out answers to complex problems like the most efficient air routes between Chicago and Singapore or the best way to route phone calls around the U.S.—conundrums that companies like FedEx and AT&T have grappled with for years—in a fraction of the time that it takes conventional computers to do. Researchers have envisioned using DNA computers for several other applications, such as a way to detect changes in live systems—like cancer within the body or the spread of contaminants in a lake.
"DNA computers may be able to accomplish things that electronic computers cannot," says Len Adleman, a molecular scientist at the University of Southern California. "For example, it is very hard to conceive putting an electronic, silicon-based computer into a bacterial cell."
A team of researchers from Davidson and Missouri Western State University in St. Joseph inserted a free-standing, circular piece of DNA (called a plasmid) into a benign strain of the single-celled intestinal bacteria Escherichia coli, some strains of which can cause food poisoning. The team modeled a simple two-pancake flip problem using two segments of DNA—one large and one short, which were inserted into the cell in a random order and orientation. The scientists also added an enzyme from the Salmonella bacteria that is capable of flipping genetic fragments.
The segments would require a certain number of flips in a given amount of time to place them in the correct configuration. Arriving at the correct orientation has a reward for each germ: immunity from the antibiotic, tetracycline. After a set period of time, the little computers were exposed to the antibiotic—only the ones with the proper segment orientation survived. From this, the researchers could tell which cells had correctly solved the flipping problem, because the ones who didn't died.
The promise of DNA computers in cells lives in the opportunity for parallel processing: Because cells are alive and replicate, copying the plasmid segments and the Salmonella enzyme into new cells, the number of individual processors working on a problem continuously multiplies, potentially allowing them to reach a solution faster than electronic silicon-based computers, Haynes explains.
Whereas the two-flapjack version of the burnt pancake problem used in the study is relatively simple, the researchers point out that for a stack of six pancakes, the possible number of stacks is 46,080, and for 12, nearly two trillion. "If you had 11 or 12 pancakes, Haynes says, "then a conventional computer would take something on the order of months to solve the problem."