Office of Communications
Stanhope Hall, Princeton, New Jersey 08544-5264
Telephone 609-258-3601; Fax 609-258-1301
Contact: Steven Schultz (609) 258-5729
January 14, 2000
Scientists Create RNA Computer
PRINCETON, N.J. -- Princeton University researchers have developed a kind of computer that uses the biological molecule RNA to solve complex problems. The achievement marks a significant advance in molecular computing, an emerging field in which scientists are harnessing molecules such as DNA and RNA to solve certain problems more efficiently than could be done by conventional computing.
In work to be published in the Proceedings of the National Academy of Sciences, the Princeton scientists used a test tube containing 1,024 different strands of RNA to solve a simple version of the "knight problem," a chess puzzle that is representative of a class of problems that requires brute-force computing. The knight problem asks how many and where can one place knights on a chessboard so they can not attack each other. For the purposes of their experiment, the researchers restricted the board to just nine squares, so there were 512 possible combinations. Of these, the RNA computer correctly identified 43 solutions.
It also produced one incorrect response, highlighting the need to develop error-checking techniques in chemical computing.
This test-tube computer does not have any immediate applications, and it will probably never completely replace silicon technology. But it does have attractive aspects, said assistant professor of ecology and evolutionary biology Laura Landweber who led the research project in collaboration with professor of computer science Richard Lipton, and postdoctoral fellow Dirk Faulhammer and a student, Anthony Cukras.
"It begs the question, What is a computer?" said Landweber. "A computer can be an abacus, it can be many types of devices. This is really an abstraction of a computer."
One advantage, said Landweber, is that the genetic molecules DNA and RNA, which encode all the instructions for creating and running life, can store much more data in a given space than conventional memory chips. Another benefit is that, with vast numbers of genetic fragments floating in a test tube, a biomolecular computer could perform thousands or millions of calculations at the same time. It is an extreme example of parallel computing, which is a rapidly growing area of computer technology.
For example, in the knight problem, each strand of RNA represented a possible solution, but the researchers did not need to sort through each one individually; in a series of five steps, a specially targeted enzyme slashed away all the strands that did not match the requirements of a correct solution. Researchers believe that such techniques could be valuable for problems that need to be solved by trial and error, where it is cumbersome to test possible solutions one at a time.
DNA computing has attracted considerable attention from researchers since 1994 when Leonard Adleman of the University of Southern California used DNA to solve a version of an archetypal problem called the traveling salesman problem. The idea is that words written in the letters of DNA, referred to as A, T, C and G, could represent the ones and zeroes used in computer logic. Computing is accomplished by eliminating molecules whose sequences appear to be poor solutions and retaining ones that seem more promising. The output of final molecules can be read like the holes punched in an old-fashioned computer tape.
Landweber found that substituting RNA for DNA gave her more flexibility in developing a computing system. With DNA, there is a limited set of restriction enzymes - a kind of molecular scissors - so scientists may not be able to cut the molecule where they want. With RNA, Landweber's group could use just one universal enzyme that targets any part of the molecule. This aspect streamlines their approach and makes it inherently 'scalable' to larger problems.
Reporters Please Note: The Proceedings of the National Academy of Sciences has made this paper available early to coincide with the publication of related results in the January 13 issue of Nature. Copies of the paper are now available to reporters from the PNAS news office: Ph. (202) 334-2138, or email email@example.com.