DNA Computer Detects, Treats Disease

Arthur Dent '99 writes "According to this article at Reuters, Israeli scientists at the Weizmann Institute have developed a DNA computer which can automatically detect and treat prostate cancer and a form of lung cancer in laboratory experiments. Theoretically, a person could be injected with this computer, and it would detect and treat any diseased cells at the earliest stages of development, perhaps preventing the disease altogether."

Comments

May sound like a joke...

[ItMustBeEsoteric]

But wouldn't this make the concept of a computer virus horrifying?

Re:May sound like a joke...

[Goalie_Ca]

IF you think about it, this DNA computer is a virus infecting bad cells.

Re:May sound like a joke...

[Anonymous Coward]

Yes you're absolutely right. In fact according to the second page of the article, normal viruses as we know them are the "DNA computer viruses" you're talking about. They're essentially just strings of nucleotides...

When I saw the article title I thought, "Wow, DNA COMPUTER!" It's not quite a misnomer, but that name is certainly misleading. These computers don't have any hardware or non-organic components--they're basically just prearranged nucleotide sequences.

The concept is interesting nonetheless; these are computers in the sense that they function as finite automata. Anybody remember taking Computability in college? =)

Re:May sound like a joke...

[Kelz]

once it's stripped of the naughty bits you have a very powerful retrovirus, the most powerful in nature.

That we know of!

Re:I have to wonder...

[merdark]

Then you get a virus.

I hate to break it to you, but we *already* get lots of cell viruses. Like the flu for example.

Besides, the research is far away from actually doing what the headline suggests. For now, they are trying to make something that can even *detect* a particular type of diseased cell.

Other questions about safty would be, are these computers capable of reproducing? I haven't yet read the literature on this, but I'd guess no. They do not use typical virus mechanics for one, and they are nowhere close to being a complete cell. If they can't reproduce, even if they did go haywire and start destroying cells willy nilly, there would only be so many of them in your body to do so. Treatmet could easily be stopped as soon as the first hint of ill effects are noticed.

"Computer" is Misnomer

[Kevin Nichols]

Calling it a computer is a bit of a misnomer. It's a molecule that recognizes the presences of a specific signal to release a payload. That's it. It is of course a clever trick, but the word "computer" is just a device used by the science journalists to make it sound more interesting.

Why doesn't anyone here understand...

[cr0sh]

...what they have done?

If this story is true, then these researchers may have unlocked a "secret" that is incredible in scope: They have learned (in a limited manner) how to code in DNA - they have hacked nature's UTM.

Such a discovery and the applications of its use would lead to incredible things - both for good and ill! Incredible "cures" and horrific weapons all at once! Instant death and neverending life at our fingertips! In some ways - I think this may have come too soon, and will end up killing off life on this planet - we can't even agree to disagree on our religion (never mind the fact that religion is nothing more than mythology and fantasy for grown adults), instead choosing to kill ourselves over which invisible man in the sky is better!

DNA (and the attendent processes for its replication - heliocase, RNA primase, DNA polymerase, etc) is nothing more than a long UTM program "tape", where the GATC are the symbols for the program - and this "tape" controls the rest of the processes in the cell (ok, if you have followed this long - you can see I am *not* a biologist by any means - I likely have some things very incorrect).

I don't know - I may be wrong - but this just seems incredible (if true)...

Re:Why doesn't anyone here understand...

[tfoss]

What they have done, which is cool, clever and generally admirable, is to add an input (detect protein A, or RNA strand B, etc.) that triggers an appropriate output (synthesise protein C, or make enzyme D to release drug E).

While I agree what they've done is cool and clever, your comment (as well as the linked article, and even the paper itself) are somewhat overstating the actual accomplishment. The original Nature paper this refers to is pretty confusing as it really tries to keep the computer analogy up throughout the whole thing. As best as I can decipher through a quick read-through (and IAABiochemist), they synthesized some long single-stranded DNA molecules. Period. The clever part about it was designing a sequence that, when bound to certain mRNA molecules, will present a known restriction enzyme cleavage site. The restriction enzyme cleaves at that site, and the resultant, shorter molecule can repeat this with a different mRNA molecule. Wash rinse repeat.

This system, and mind you, this is only a model system created in a test tube, free of all the myriad cellular components that might muck it up, only involves inputs and outputs that are small nucliec acids. They do nothing to synthesize, make, or create any proteins (or drugs in the typical sense). The "drug" in this case, is simply a short strand of ssDNA that can prevent the translation of a specific mRNA sequence. The fact that you can do that is, in itself, tremendously cool and potentially therapeutically useful, but is far from novel.

My beef with this it that, while in the strictest sense it might be a "computer," that is a loaded word that implies far more than this research actually delivers. It is a computer in the same sense that the door lock on your car is. It can distinguish a pre-designed set of inputs (certain mRNA sequences vs. a certain set of hills and valleys on your key), and react by either doing something (get cleaved to release a toxic DNA sequence vs. allowing you to physically turn the lock) or not. So, while a novel application of nucleic acid binding, all this talk of 'inputs, computation modules, logical control, and autonomous biomolecular computers' is mostly fluff. (Granted, to get published in Nature or Science you generally need a level of such fluff).

-Ted