Slashdot Mirror
Getting Closer To DNA Computing
jwambach writes: "This brief article over at Yahoo! describes how scientists are attempting to create molecular electronic circuits using DNA. They say that they could potentially create circuits 10,000 times smaller than with current technology." The article says that "[r]esearchers have already created molecular wires, logic gates (a building block of computers) and switches, which could be hooked up to make a working computer a fraction of the size of ones based on silicon chips." The coolest part is that assembling the tiny components could be done with DNA-tagged components, shaken (or stirred) in a test tube.
New form of artificial intelligence. DNA evolves with useage. Computer refuses to run Windows and won't take input from user attached to keyboard.
So can I take these DNA circuits and introduce them to my body.. Can I overclock my brain? Can I change my mind? Johnny Mnemonic here we come!
air and light and time and space
The article suggests that we have all the molecular tools and that it is a straightforward matter of hooking them together to make molecular scale computers/robots....this is misleading. There have been a number of published papers that describe molecular logic gates (based on 'inputs' of ions, or particular DNA strands) and the 'output' is often flouresence...which doens't allow you to hook these components up into large arrays. I have tones of ref's if anyone cares to look them up. Even things like Adleman's approach, while actually quite brilliant in design, fail to provide either faster more efficient computers or application to control (bio) chemical systems. Winfree's stuff with Seaman's DNA cubes is again quite interesting, but it's difficult to see where it will go or what insight it will provide. Anyone remember bacteria rhodopsin based memories? I think work continues on them, but no one sees practical application. The stuff coming from UCLA and HP (rotaxanes) is actually quite interesting; and given the population of things that has to do with molecular computing; it probably has the greatest chance of actualy realization.
Most real nanotech you see today is just some real nice bio/organic/supramolecular chemistry that has been moving along since the inception of chemistry (a-hemolysin stuff, self replicating peptides, nanotubes etc.). Not to say that it isn't wonderful work or un-useful...but it is just renamed chemistry.
The other thing I have a problem with people like drexler is that suppose one does actually succeed in producing one of the molecular machines he describes through synthesis or AFM techniques...we don't have the technology to characterize such beasts. Ask an enzymoligst how many molecules of enzyme he needs and the things he neeeds to know before he can say what an enzyme is doing (or what it looks like). Enzymes are in many ways much simpler than the kinds of things Drexler proposes. In other words even if you could build a nano-bot; I don't know how you could know it was actually there or convince other scientists given the trouble we have with the molecules that nature has given us.
What is frusturating out the whole molecular computing field is that when you actually try to sit down and do it, it seems impossible to try to figure out a way to do computation on a molecular level....yet natural things like remarkably organized biological cell and the human brain show us that there is a way, but is it the only one?
I knew I had read about a very similar set of experiments going on with DNA in Wired. I didn't realize how long ago it was. Anyhow, I did me a search over there on the mag and found the article in their archives called Gene Genie -Aug1995. This is a rather old article, and I honestly have no idea how far this guy got with his research.
To quickly summarize, this Adleman fella actually got DNA to do some simple calculations. What the biggest stumbling block at that time (and most likely still) was getting an interface between the user and the DNA setup. Sounds like that's what these folks in England are up to now.
I feel I should clarify one point that seems to be getting mixed up here on this thread. The promise of DNA based computers isn't specifically to replace CPU's that we know today. In fact, DNA is far slower at running the repititive kinds of tasks that silicon deals with. The exciting stuff is in it's ability to do massive parallel processing on a small scale.
For example, if you had a race between silicon and DNA to count to 1e24 the silicon would most likely be a clear winner. On the other hand, DNA based processors usher in the ability to solve problems that silicon based machines simply can't do, like calculate a perfect game of chess.
Most likely, what we'll eventually see out of all this is a new hybrid computer utilizing both silicon and DNA based on their individual strengths. The prospects for this are both very cool, and very scary. We're talking about a machine that literally removes every advantage the human brain presently enjoys over silicon CPU's in the way of problem solving. Imagine HAL, only one hell of a lot smarter.
The line must be drawn here. This far. No further.
To clarify a few things:
There is a big difference between quantum computing and DNA computing. In fact, there is a big difference between what people usually mean by DNA computing, and the results described here in this article. To summarize (and I'm condensing things a bit for brevity):
Quantum computers use quantum states instead of discrete states to perform computations. Quantum bits or "qubits" are quantum superpositions of the "pure" states corresponding to 0 and 1. (e.g. if we consider the 0 and 1 states to be two different energy levels of a certain electron in an atom, if we don't observe the electron, we can place the electron into a superposition of states corresponding to the two energy levels). The power of quantum computing (in theory, at least) comes from using these qubits instead of normal bits because by acting on a superposition of quantum states, we perform the action on each quantum state in the superposition individually.
Quantum computers actually require a completely different model of computation than the one that current computers are based on, and are potentially (if all this theory pans out) counterexamples to the Church-Turing thesis (that all computing devices are polynomially equivalent to a Turing machine).
DNA computing is more about mass parallelization than a new model of computing. It is based on the idea that by encoding solutions of combinatorial problems (read: NP-Complete) in DNA, and using modern DNA manipulation techniques (which include searching for particular sequences), we can find a solution with particular properties (read: optimal). For example, to solve the Travelling-Salesman Problem, we can generate (randomly) trillions of solutions and then pick out the optimal one. The advantage is that generating these is quick, and any operation you perform is performed on each strand at the same time. This massive parallelism makes it possible to solve search problems (TSP, Satisfiability, etc) that were not previously feasible, but DNA computing still falls under the Church-Turing thesis and so all the same complexity-theory results apply. DNA computing just reduces the constant involved by a factor of 100000000000000 or so.
The article mentioned here seems to point more towards a "computer built with DNA", where DNA is being used to build components that will mimic the action of a circuit-based computer. It seems that this would simply bring several orders of magnitude of miniaturization to current computers, rather than anything fancy-schmancy in the theoretical computer science realm.
Donny
P.S. If you saw something on work at Los Alamos, it's probably on quantum computers.
We all consider cooling to be of vital importance in our silicon based machines, don't we? What kind of new cooling device are they going to create/use that will keep the DNA from degrading or just plain denaturing? Not that this isn't an interesting experiment, but the actual implementation seems rather difficult and/or unstable.
1. The article does not talk about computing by DNA, it's about using DNA to build smaller conventional electronical circuits. So the logic would not change.
2. As devices get smaller, required amount of electricity drops. Future pc's/appliances should be able to run on bare light/movement energy. "Cool."
3. Imagine tools for DNA design. Anyone could design their own special component/appliance and start a farm. Back to the nature!-)
I think, therefore thoughts exist. Ego is just an impression.
I think there's a basic misunderstanding going on here. It might be too late to get it moderated up, but here goes:
;] ). These DNA-based computer's don't have all the cruft we usually associate with DNA. For example, these things won't reside in cells. The individual gates, transistors, etc., are just carefully ordered atoms. There's no basic difference, other than the fact that we can control, very carefully, how these atoms are ordered. We can't do that with silicon. At least, not to this degree. We're using nature's own techniques to control how these DNA-based(not actually the life-deciding chemicals, just based on them) chemicals to an incredible degree. That's how come they'll be so small.
Most people here seem to have the idea in their head that these DNA-based chips will degrade and/or mutate. Don't think of these as life forms, but think of them as life-inspired. Your silicone chips are nothing more than carefully ordered atoms, which are called "chemicals"(
As for the degradation worries, there's really relatively little to worry about. I'm not a biologist, but these people won't be building these things using unstable molecule chains. They can't, really. They'll have to make them fairly stable. As for the "mutation", it's no different than in silicon. Sure, a stray neutrino isn't going to affect a relatively MASSIVE silicon-based gate, but if you made that gate REALLY small, then yeah, something really small could permanently damage it. It isn't that DNA-based computers are more vulnerable because they're DNA-based. They're more vulnerable because the individual parts are smaller. If you shrunk your processor to the point where the transistors were as small(and had as few atoms) as a DNA-based transistor, then you'd have the same problem. Very little upsets would completely destroy such a silicon transistor.
Just to recap:
These computer's won't be using the same stuff in our cells. They'll just be using strings of atoms, carefully ordered(much like silicon is carefully ordered, only in this case we can be very precise).
The worries expressed here often(that of mutation and degredation) are really besides the point. They won't build these things with unstable molecules that "melt" at room temperature(well, maybe in the beginning, but that'll have to change), so a chip decaying is not really likely. And the problem with mutation(where something happens to an individual component) is a risk involved in any device which uses extremely small parts.
Dave
Barclay family motto:
Aut agere aut mori.
(Either action or death.)