Len Adleman: Tapping DNA Power for Computers

This math professor is convinced that computing with the molecular structure of genes will someday whip silicon-based machines

Len Adleman is a university scholar with Hollywood appeal. Last year, researchers from Steven Spielberg's studio called the University of Southern California mathematics professor to chat about something called DNA computing. Apparently, the legendary director was contemplating using Adleman's groundbreaking work in the nascent field of substituting DNA for silicon as the basis for computing in a thriller starring Tom Cruise.

It's no wonder that the kings of make-believe wanted to take a meeting. Adleman has strong comptuing credentials. His name stands for the "A" in RSA Data Security, an algorithm that has become the de facto standard for industrial-strength encryption of data sent over the Internet. Moreover, the ideas he articulated on using the molecular structure of DNA to process information could one day result in computers many times faster and more powerful than existing machines. Ultimately, engineers hope to weave these miniature DNA computers into the human body to help monitor and prevent disease.

A soft-spoken 56-year-old with floppy brown hair, Adleman happened upon the concept of DNA computing by accident. In the early 1990s, he was working hard to evangelize a mathematical theory he believed could provide a cure for AIDS. He had realized that as certain cells were depleted, other cells -- similar in type but not in function -- proliferated proportionately. Adleman reasoned that if a population of an unaffected cell type, such as T-8s, could be artificially reduced, the immune system might naturally increase production of the T-4 cells that the HIV virus depletes.


  "It was a period of great frustration. Lots of people were dying...and yet I was meeting with only modest success," remembers Adleman. His AIDS theory hit a dead end but not before Adleman decided to learn molecular biology so he "could better speak the language of the powers that be in the AIDS research community."

After a year in the lab, he realized that strands of DNA behave much like mathematical equations. DNA's chemical bases -- adenine, thymine, cytosine, and guanine -- hook up in a predictable manner: adenine always links with thymine, and cytosine with guanine. Because of the regularity of the pattern, Adleman hypothesized that he could use molecules to process data the same way PCs use microprocessors.

He decided to test his theory by seeing if a test tube of DNA could solve the Hamiltonian Path problem, also known as the Traveling Salesman problem. The idea is to find all possible paths between a certain number of cities without visiting any city more than once. The problem, a classic mathematical conundrum, becomes exponentially more difficult as more cities are added.


  In the lab, Adleman made a strand of DNA to represent each city and the path between each city. He then encoded the sequences so that a strand representing a road would connect to any two strands representing a city according to the rules of DNA binding. Then he mixed trillions of copies of each strand in a test tube. Within seconds, the strands wove themselves together in a myriad of possible combinations.

Over a period of time, Adleman performed a series of biochemical reactions to eliminate the wrong answers -- strands encoding routes that either started or ended in the wrong city, those that visited a city more than once, and so on. When all the wrong answers had been destroyed, Adleman was able to look under the microscope and find only strands that carried the right answer.

Adleman's experiment used just seven cities, a problem that isn't hard to solve on modern computers. But Adleman's biological computation showed that DNA has the potential to solve far more complex problems than even the most advanced electronic computers can. The fastest supercomputer wouldn't be able to solve a problem with more than about 50 cities, Adleman says. He believes that a test tube full of DNA could solve the problem with as many as 200 cities.

Here's why. First, DNA is incredibly enery-efficient. Adleman likes to point out that billions of years of evolution have pushed cells to the brink of what thermodynamics says is possible. Take ligase, a molecule whose job it is to stick strands of DNA together. With just one joule of energy -- the amount of energy a human expends to lift one kilogram one meter, ligase molecules can perform 20x10 to the 18th operations, Adleman says. That's a million times a million times a million times 20 operations. Such efficiency could push computing to new levels since electronics are limited by the amount of power -- and the heat it gives off -- needed to run increasingly sophisticated operations.


  DNA is also a wonderful way to store information. One gram of genetic material, which would occupy about one cubic centimeter, can hold as much information as 1 trillion CDs, according to Adleman. It's also incredibly cheap: Commercial labs sell a molecule of DNA for about one-thousand-trillionth of a cent. The cost is about $30 for a DNA sequence big enough to compute on. Intel sells its latest P4 chip for more than $500. "DNA has been storing the blueprint of life for several billion years," says Adleman. "Its powers are an untapped legacy for the 21st century."

Since 1994, dozens of research groups around the world have jumped into the emerging field. Erik Winfree, a 32-year-old associate professor at the California Institute of Technology, was studying robotics when he first read about Adleman's DNA experiment. Within four months, he had changed tracks to focus on molecular computation.

In 2000, Winfree won the prized $500,000 MacArthur Fellowship for his work that demonstrated that it's possible to create nanoscopic -- really, really small -- building blocks out of DNA that can both store data and be "programmed" to do mathematical computations. This implies that biological cells are themselves computational devices and that understanding how cells regulate genes may depend on identifying the calculations that they perform.


  The first practical applications are also emerging. Last January, Olympus Optical Co. and the University of Tokyo claimed to have jointly developed a fast way of identifying genes associated with diseases. Researchers developed a process that synthesizes 10,000 different DNA strands that are known to bond with genes related to specific diseases such as cancer, Japan's Nikkei Weekly has reported.

The strands are numbered and mixed with fluid containing genes extracted from the patient. The fluid is then tested to determine which genes are functioning in the patient's cells by reading the number of the DNA strands that appear after a series of biochemical reactions. Researchers claim they can complete a single test in about three hours, about one-half to one-third the time taken by conventional biological methods.

Don't throw out your old PC yet, though. Adleman says it will be years -- if ever -- before test tubes of DNA replace the common computer. While he has proven that DNA can be used to calculate complex mathematical problems, the work remains incredibly time-intensive. Take the Traveling Salesman problem. The DNA in the test tube produced 100 trillion answers in less than one second. Most of those answers were repeats -- and incorrect. Adleman had to discard the erroneous answers using lab procedures that took about a week.


  Thus, as the complexity of problems increases, the manual labor required for DNA computing could outweigh the benefits of superfast computation. "Here we have the most amazing tool chest we've ever seen. We know it's great because it was used to build you and me," enthuses Adleman. "Right now, though, we are very clumsy with it."

Most scientists today believe that DNA computing will complement silicon-based computers, rather than replace them. And that's fine with Adleman. "I think the best thing about DNA computing is that it's helping create a new generation of scientists who understand biology and mathematics simultaneously," Adleman says. He hopes his work can ignite a new age of science -- akin to that in which interdisciplinary scientists like Leonardo DaVinci and Galileo thrived.

By and large, that's heresy in today's world of science, where specialization is king. But just as the Renaissance geniuses ultimately proved prescient, so could Len Adleman.

By Jane Black in New York

Edited by Alex Salkever

Before it's here, it's on the Bloomberg Terminal.