By Avi Machlis
Small Times Correspondent
REHOVOT, Israel, Dec. 18, 2001 — For a scientist who has just staked a claim to the first programmable and autonomous biological nanocomputer, Professor Ehud Shapiro is remarkably low-key when asked to predict how such research may eventually change the world.
He refuses to get drawn into detailed discussions of
 |
| Ehud Shapiro’s DNA computer encode zeroes and ones in the an input molecule with an exposed “sticky” end. Then, another DNA strand — the software — swoops in to try and hook up with an exposed edge (upper left). After hooking up, an enzyme called ligase (upper right) seals the link, and another called Fok-1 moves in to snip the strand (lower left), leaving the next section exposed. The process continues several times until the computer delivers an answer to a question. An “output detector” DNA molecule (lower right) then binds to the resulting output sequence. |
Shapiro does not see his computer as a potential competitor to silicon-based electronic computing, as some have suggested. Instead, he envisions DNA computers as a “molecular computing device that can operate initially in a test tube and eventually inside an organism and interact with its biochemical environment.”
DNA computing could possibly be used to streamline laboratory analysis of DNA, by eliminating the need for sequencing. This, he said, could happen within a decade.
“In the longer term, you may have medical applications in which this device can operate in vivo, inside a living organism,” he says. “Based on the information it receives from the environment and medical knowledge encoded in the software it may diagnose the problem and prescribe a solution, and then it could synthesis that molecule and output it.”
That’s as far as Shapiro is willing to venture on the prospects of the technology.
“I don’t have an opinion on nanogurus or nanoapproaches,” he says dryly during an interview in his office at the Weizmann Institute of Science in Rehovot, Israel. “We know where we are and where we are going to go. It’s just going to be a very long way.”
The starting point for Shapiro, who recently published his design for a molecular computer in Nature magazine, came after his Internet software company called Ubique was sold to IBM in 1998.
Plotting a path back to academia, Shapiro stumbled upon research being done in molecular computing, and challenged Yaakov Benenson, a biochemistry Ph.D. student, to help make it work. Their modest initial goal was to find a way to use turn DNA into the most elementary mathematical computing device known as a finite automaton, capable of answering “yes” or “no” to very basic questions about a bunch of zeroes and ones.
“We constructed a molecular realization of this mathematical device,” Shapiro says. “It has input, it has software and it has hardware components; and when it computes it produces output, which is another molecule.”
To do this, Shapiro and his colleagues used the four components of a DNA strand known as A, C, G and T to encode the zeroes and ones and create an input molecule with an exposed “sticky” end. Then, another DNA strand — the software — swoops in to try and hook up with an exposed edge like a Lego piece attempting to lock into a complementary block. Each exposed edge has a specific complementary DNA strand.
After hooking up, the hardware gets to work. An enzyme called ligase seals the link, and another called Fok-1 moves in to snip the strand, leaving the next section exposed.
The process continues several times until the computer delivers an answer to the question. There are 765 different possible software programs that can be used for simple calculations, such as whether there are an even or odd number of zeroes or ones.
Shapiro’s research is the latest step forward in a field founded by Leonard Adleman of the University of Southern California, Los Angeles. In 1994, Adleman proved that DNA could compute, when he used the stuff to solve the “traveling salesman” problem, in which the shortest route between several cities must be mapped without going through the same city twice.
Conventional computers have extreme difficulty solving the problem, especially when dealing with many points on a map. This is because electronic computers are based on sequential logic, which makes them good at solving a problem requiring lots of computations in a row. But posed with a puzzle of how to figure out the shortest route between 100 cities — a problem best cracked by simultaneously performing an enormous number of short operations — conventional computers do not make the grade. Adleman demonstrated that DNA could be an efficient way to solve such problems.
Shapiro says his DNA computer is fundamentally different from Adleman’s breakthrough. Although Adleman’s computer was composed of many trillions of tiny DNA molecules swimming around in a test tube, Shapiro says it was essentially a large operation that required active involvement of scientists.
“The calculation needed to be carried out by humans. In our case, the computer is just the molecules,” says Shapiro, who can put a trillion of his own biological computers into a drop of solution. “His computer is measured in meters, ours is measured in nanometers.”
Experts point out that Shapiro faces stiff competition and will be challenged to scale up the work to perform more complex computations.
John Reif, professor of computer science at Duke University, described Shapiro’s work as “ingeniously constructed experiments” that clearly demonstrated the ability to perform simple computations via solid experimental protocols.
“But there is a lot of competition out there in the DNA computing world,” he added, singling out DNA computing research at Princeton University and the University of Wisconsin that has gone beyond the finite automaton.
“People are really aggressively pushing the limits, so the challenge for the Israelis is to go in and push those limits as defined by some of those strong competitors,” Reif said.
Shapiro has no illusions. The biggest stumbling block now is the dependency on natural enzymes, meaning scientists must search for the right enzymes that could help perform computations on DNA. Science still has no clue how to create designer enzymes that could pave the way to dramatic progress.
For his part, alongside the finite automaton, Shapiro has taken an important theoretical step forward by building a model of a molecular Turing Machine, which is a representation of a computing device capable of an infinite number of computations. It is in this green, squarish model, sitting in a cardboard box in his office, that Shapiro sees the real potential for molecular computing. The ability to create a molecular Turing Machine would allow scientists to use DNA to generate massive computing power.
In the meantime, he is keeping focused on the scientific challenges ahead — and plans to be tied up in his DNA strands for a while. “We have made a first small step in this direction,” he says. “I believe this will keep me busy until I retire.”
Related Link
See a visual presentation of Shapiro’s biomolecular computer.