While the rest of us surf the Internet, Russell Deaton explores ways to store its contents in a drop of liquid.
Deaton, professor of computer science and computer engineering, along with his University of Arkansas College of Engineering collaborators Steve Tung and Jin-Woo Kim, and Junghuei Chen, a biochemist from the University of Delaware, is investigating methods of storing biological and non-biological information in a memory made of DNA (deoxyribonucleic acid) molecules.
Essentially, they’re developing a DNA computer.
The group is in the third year of a three-year, $699,000 grant from the National Science Foundation. Deaton admits the concept is on the edge, but the more he explains it, the more natural the whole idea seems.
“Engineers build tools to help humans do math and solve problems, and a computer is just another tool that we’ve built and programmed to perform complex computations,” he said. “We’ve always used natural materials to build these tools. For example, an abacus was built out of wood. And a computer is made of silicon, which is the primary component of the earth’s crust. In our case, we’re building a tool out of material that just happens to be biological.”
Further expanding computer technology, Leonard Adleman, often called the inventor of DNA computing, published the first article connecting DNA and computations in the journal Science in 1994. “To build a computer, only two things are really necessary; a method of storing information and a few simple operations for acting on that information,” he wrote.
So eventually instead of laptops we’ll carry test tubes, right?
Wrong, according to Deaton. “DNA computers probably won’t ever replace solid-state computers, which are faster, more reliable and more predictable,” he said. “But you might use DNA as a device to store an enormous amount of data. For example, you might be able to carry around the entire IRS database.”
Great, let’s go. Can’t wait to buy the newest gadget; a DNA memory device.
Slow down, techster, they’re not going to be at Best Buy for a while. “Consider what we know about physics today, and we’re Newton with his apple,” Deaton said, shaking his head. “Modern electricity? We’re Ben Franklin with a kite. Cars? We’re not even Henry Ford, we’re inventing the wheel.
“It’s a little frustrating, but how often do you get the chance to play a part in the founding of a whole new field?”
A New Twist on the Double Helix
Why are roughly 100 to 150 researchers in the world establishing the field of DNA computing at all? What’s the lure? Most important, what’s in it for us, the impatient computing public?
The most pressing reason to search for alternative methods of processing information is the sheer volume of the information we’re processing. “Humans produce about 1018 bytes of information every year, which completely boggles my mind,” said Deaton. “That includes books, newspapers, magazines, Web sites, blogs; you name it. We have to find better ways to deal with this information avalanche; to store it, search it and retrieve it.”
And DNA computing promises the potential of much better ways to do exactly that.
The Power of Density
The possibility of storing vast amounts of data in a small space is probably the most alluring benefit of DNA computing. Think of how much information is stored in the individual DNA of every human. After all, you’re unique, just like the other six billion people on the planet.
So how much non-biological data might be stored? “One DNA sequence might represent one byte of information,” said Deaton. “You can store potentially an exabyte (1018) of information in a drop of water. The sheer density of what you can store is exciting, to say the least.”
Adleman puts it another way when he writes, “One gram of DNA, which when dry would occupy a volume of approximately one cubic centimeter, can store as much information as approximately one trillion CDs.”
Indeed, the entire contents of the Internet actually could be stored in a drop of water.
The Power of Energy
With great technological marvels have come great demands on the electrical supply. DNA computing offers the hope of a remarkably energy efficient machine, more than a million times more efficient than a PC.
Adleman writes that in principle, one joule is sufficient for approximately 2 x 1019 ligation operations using DNA. Existing supercomputers now execute, at most, 109 operations per joule.
The Power of Endurance
Remember the last crime show you watched? The detectives probably used a tiny strand of hair or even an old sample of DNA to crack the case.
“Scientists like to say that DNA is robust,” said Tung, assistant professor of mechanical engineering. “You can extract it from old bones and still test it. That toughness makes it very appealing from a materials standpoint.”
The Power of Parallel Processing
Atraditional computer stores information as sequences of zeros and ones in memory, then manipulates that information with the operations available on the processor chip. Because most electronic computers operate on a sequential basis, they essentially perform tasks one at a time, albeit very quickly. DNA computers, however, offer a much more intriguing reality: massively parallel processing.
In the cell, DNA is biochemically modified by a variety of enzymes, tiny protein machines that read and process DNA, manipulating it on the molecular level. Some enzymes, for example, cut DNA, and others paste it back together. Other enzymes function as copiers or repair units. We can now perform many of these cellular functions in the test tube. And in the test tube, these enzymes can work on many DNA molecules simultaneously.
Which means DNA strands can produce billions of potential answers simultaneously, making DNA computers incredibly well-suited for solving problems that require searching for solutions among a massive number of possibilities.
Imagine searching the Internet with one click, looking based on content instead of address. The research group also is exploring massive DNA memories that store and search information based upon meaning, or semantics.
“The operations on DNA molecules occur in parallel, so searching and retrieving data based upon both content as well as context is possible,” said Chen, associate professor of biochemistry at the University of Delaware.
Memories with Meaning
A human DNA molecule is about a meter long and a twenty-millionth meter wide, or the width of 20 hydrogen atoms. The famous double helix shape is like a twisted ladder, and each rung is made up of four nucleotides adenine (A), thymine (T), cytosine (C) or guanine (G). The DNA code is then expressed in combinations of those letters.
By mapping data onto sequences of nucleotides, information is stored in DNA. That information can then be retrieved by placing the Watson-Crick complement in the test tube. The Watson-Crick complement which replaces A with T and G with C then matches with the sequence representing the information you want. This base-pairing, template-matching reaction, which pairs up Watson-Crick complementary sequences, is called DNA hybridization.
“The DNA molecules are floating around and they want to pair up,” said Deaton. “Our goal is to make those DNA sequences have meaning. We could build a DNA dictionary of words to create a library of DNA sequences that would store information. Then we’d use the base-pairing mechanism to do a search.”
If the team can create a library of sequences that’s controllable, predictable and well-behaved, then instead of conducting an Internet search by address, you could eventually conduct an Internet search by both context and content.
Take these two sentences:
“I sat on the bank of the river.”
“I cashed my check at the bank.”
They could be stored in DNA by having each word represented by a DNA sequence. For example, “bank” could be “AGATGC,” and “river” could be “CCGTGA.” In the test tube, these DNA words could be combined into one molecule representing the sentence. So all sentences with the word “bank” could be retrieved by hybridizing with its complement “TCTACG.”
But you may have noticed the word “bank” has two different meanings in these sentences. Which meaning is which is determined by the context, or the other words in the sentence. “We can search for the first sentence and extract out all banks associated with checks, for example,” said Deaton. “Or use the other context and retrieve banks associated with rivers by biochemically searching for molecules that contain the relevant DNA sequences.”
So DNA memory doesn’t work like the memory in your typical desktop computer, which stores information at specific addresses. Information is instead retrieved by content, or the DNA sequences that represent the information.
Of course, the DNA sequences that are used have to be designed to allow reliable and error-free information processing. “It always comes down to a word design problem,” Deaton said. “In DNA computing, things that are not a problem for biologists are definitely a problem for us,” he said. “Like achieving the required level of control and reliability over what’s going on in the test tube.”
Enter Jin-Woo Kim, assistant professor of biological engineering, who looks for the solution within solutions.
“I implement computation in the test tube given many different assumptions and conditions,” he said. “Writing code is a prediction. I simulate and verify that prediction under real biological conditions.”
That verification is vital because, unlike genetic DNA, the biotic DNA used for computing does not contain enzymes to automatically correct mistakes. “Each sequence might contain thousands of A-T-C-G base pairs,” said Kim. “One change in one base pair makes the entire outcome totally different.”
Ensuring that the information in the DNA solution is absolutely correct is simply the starting point of Kim’s work. “How can we see what’s going on in the test tube?” asks Kim.
The current method, gel electrophoresis, involves placing a solution of heterogenous DNA molecules on one end of a slab of gel and applying a current. The negatively-charged DNA molecules move toward the anode. Because shorter strands move quicker than longer ones, the DNA is separated by length. Using fluorescent dyes, a chemical wash and ultraviolet light, you can see bands in the gel where the DNA molecules of different lengths have come to rest.
Kim also uses microarray technology, which involves putting a DNA sample on a slide, then adding another, using a chemical wash and then a high-resolution scanner to determine what’s happening. But this method has many disadvantages, as well.
“It requires a very sophisticated, expensive and large machine,” said Kim.
Kim, working with Tung and Deaton, is trying to discover a way to detect the DNA reaction in real-time. “We’re working on a highly-sensitive, electrically-addressable, real-time, DNA-based carbon nanotube-wire sensor. Within five years, we’ll have something that we can use.”
Steve Tung, assistant professor of mechanical engineering, spends his days building nanomachines and making them work. “My part of this research is building a platform for a DNA computer to operate,” he said. In other words, he’s developing a nanomachine to retrieve the information the other researchers are storing and searching.
How big is a nanomachine, anyway?
If you can spare one, pull a hair from your head (hey, no one ever said science didn’t hurt). A nanometer is about a thousandth of the width of your hair, or one billionth of a meter. Put another popular way, if we measured the earth in nanometers instead of meters, it would fit in a shot glass.
“A nanotube generally has a diameter ranging from one to tens of nanometers, and its length is of the order of one micrometer,” said Tung. “An exact size depends on the individual nanotube.”
Weird things happen with materials that small.
“Conventional methods just don’t work at that level,” said Tung. “Materials with certain properties take on totally different characteristics. For example, we can’t predict how molecules will align themselves, how they’ll react with other molecules or whether they’ll stick together.”
This stickiness issue is quite, well, sticky. “The smaller the size, the more things stick together,” Tung explains. “Think of a golf ball on your car. If you drive away, it will come off. But think of particles of dust on your car. They stick to it.
“As it is, DNA exists in a mixture or in suspension and it’s hard to manipulate. We’re working on chemically binding the DNA onto the carbon nanotubes,” said Tung. “The simplest way to explain our device is that we use the change in electronic properties of nanotubes to monitor the behavior of DNA. First, we record the electronic properties of ‘bare’ nanotubes. Then, we attach single-stranded DNA to the nanotubes and record any change in the electronic properties of the nanotubes. Finally, we attach ‘matching’ single-stranded DNAs to the hybridized DNAs that are on the nanotubes and record the electronic properties again. Based on these results, we can build a hybrid nanotube/DNA device and use it to detect the presence of matching DNAs in a medium by monitoring the electronic properties of the nanotubes.”
Whew. But one wonders a rather basic question: How do you engineer a machine too tiny to see without a microscope?
So far, there are basically two processes. The “top down” methods take a larger piece of material and make it smaller. The “bottom up” techniques involve building the material atom by atom. The two approaches carry over to mixing molecules. “You can mix it as is, which is like putting marbles in milk and stirring,” said Tung. “Or you can be more precise and direct heat to change the direction of the polymers.”
But why use carbon nanotubes?
“Carbon is inert and it’s solid,” said Tung. “It’s the vanilla of all the elements. You can change it to strawberry or tutti-frutti or rocky road.
“In the ’80s, the computer revolution hit the world. In the ’90s, the Internet changed everything,” said Tung. “I believe the groundbreaking technology in 10 years could be carbon nanotubes.”
“Wouldn’t it be neat to design a computer out of a living thing?” thought Deaton as an undergrad in biology class. His interest was renewed after he read Adleman’s groundbreaking article as a professor. A recent $100,000 NSF grant, with Kim and John Lusth, associate professor, computer science and computer engineering, will help him pass along that interest to undergraduate students in the College of Engineering at the University of Arkansas.
“Undergraduate programming texts are pretty dry,” said Deaton. “I think most students would be more excited to learn those basic skills while doing work for DNA computing.”
As amazing as the concept of DNA computing is, Kim remains most amazed by the elegance of DNA itself. “Here is the basis for our bodies, the basis for everything; the source of life itself,” said Kim. “Yet the material is so simple. And there is so much potential and so many more mysteries.”
Exploring those mysteries excites Tung, who sees a bit of irony in his work. “Whatever we do, we’re trying to mimic nature, but on a much smaller scale. So you can’t predict what’s going to happen in the lab. And not knowing is the fun of research.”
Research as fun? Most definitely, according to Deaton, who has an undergraduate degree in English and advanced degrees in electrical engineering. “I’m learning new things every day, which are not necessarily in my area of expertise. And I told myself that if I was going to be in academia, I was going to make sure that I would work on something that was challenging and fun. DNA computing is fun.”