That sounds awesome, but I don't understand much of DNA, so I can't tell, if this is really a breakthrough, or rather a good step forward ... (I suspect the latter)
Something I learned recently is that DNA forms so called "gene regulatory networks" that are like neural networks. Each gene has typed chemical inputs and outputs, acting like an individual neuron. That means a cell can execute arbitrary programs. One such program is that which differentiates stem cells into organs, creating the human body in all its complexity. But we could take over this mechanism and use cells to produce what we want. We'd have self replicating factories.
This is just one way of getting to "programmable cells." Basically what they're doing is creating a database of DNA sequences that can be stringed together to create programs in cells. There's even GUI editors for this stuff, tons of standards around the syntax, etc.
But Biology is messy. This stuff isn't going to be running deterministically, it requires a deep understanding of the organism, it requires a deep understanding of the overall environment an organism operates in, etc.
As a software dev, who watches an unnecessary number of these cell internals/molecular dynamics/spontaneous self assembly vids on youtube for kicks, I am always amazed by the utter chaos that produces all these mechanisms and processes.
We see nothing like this level of randomness in the already chaotic world of software development. When one works on complex projects with more than a few hundred people one constantly sees random things bumping into each other producing stuff nobody imagined. But this stuff happening inside a cell is just insane too me. Makes me feel like a cave man playing with my current set of stone age tools.
It IS interesting in that this appears to be general (judging by the quotation at the bottom, I haven't read the paper yet, hopefully its not behind a paywall). Back in college I read about and did a project on 'DNA computing' based on some of the early results (late 90s). Back then each instance of 'computation' had to be setup manually and very specifically engineered to the exact problem at hand. I believe they solved a Hamiltonian Path problem with 5 or 6 nodes or similar. Something they could tractably check. But the actual period of 'computation' consisted of shaking a test tube a couple seconds. Then they had to sequence the DNA to get the answer out... but hey, it was early days. The book supposed that sort of thing was a fundamental limitation of the model, that it might be usable for really difficult problems but would always have to be specifically engineered and run as a one-off experiment to get the result.
This press release drove me crazy the way it was written, both insulting and deceptive, but the research sounds very interesting. They mention doing arithmetic, which would make it pretty general. I'd love to know if they have any means of doing loops. There's mention of Turing machines in the press release but I'd guess that's just something that popped up in the weird garbage can of the authors mind while grunting the thing out.
Something like this combined with the ideas of 'amorphous computing' (think tons and tons of tiny unreliable processors communicating over short-range links, computation happening primarily as a large-scale emergent property of the whole system) might help things get truly weird in the future... thinking RNA paint? Sure, OK.
I'm not an RNA guy, but as far as I can tell, the highlight from the paper is that they've shown that you can computationally design these networks in an RNA-based system.
>> I guess the somewhat interesting question is, 20+ years from now, will biological computers be DNA or RNA based?
What a world! There is far more in your sentence than the average geek might understand if she reads it too quickly. It could be nice to help everyone understand while keeping the humouristic touch.
Reaction to headline: Now program them like computers to compute like computers. I want an exponentially growing CPU!
Reaction to article: WTF is this garbage? We've been able to build RNA sequences forever. Anyone who has read ANYTHING about bioinformatics knows RNA works like 'computer code' to tell a cell how to make a protein. This is high school level biology.
'The researchers made their invention by first modelling all possible RNA sequence interactions on a computer.' No, they didn't. Jesus Christ no they didn't. I don't know what they actually DID, or what they even CLAIM to have done, but they did not do this. First off, there is no limit to the number of possible RNA sequences. Secondly, there is no limit to the variety of possible interactions. You know Folding@Home? It simulates protein folding? That's what happens after you feed some RNA into a cells protein factory. So even if you figured out the infinite variety of proteins possibly created by the infinite variety of RNA sequences interacting with themselves and all the rest of the cellular contents (including potential viruses or bacteria that snuck in), then you'd have to solve every single folding problem on all of them.
Once you get to the actual words of the actual people doing the science, hey magic, it actually makes sense and sounds cool! But WOW, whoever wrote that press release up to the point where the closing quotation starts? That person needs to never write a press release for a scientific finding ever again. Humanity will be better off!
12 comments
[ 3.7 ms ] story [ 37.7 ms ] threadLook into synthetic biology and biological parts: http://parts.igem.org/Main_Page
This is just one way of getting to "programmable cells." Basically what they're doing is creating a database of DNA sequences that can be stringed together to create programs in cells. There's even GUI editors for this stuff, tons of standards around the syntax, etc.
But Biology is messy. This stuff isn't going to be running deterministically, it requires a deep understanding of the organism, it requires a deep understanding of the overall environment an organism operates in, etc.
We see nothing like this level of randomness in the already chaotic world of software development. When one works on complex projects with more than a few hundred people one constantly sees random things bumping into each other producing stuff nobody imagined. But this stuff happening inside a cell is just insane too me. Makes me feel like a cave man playing with my current set of stone age tools.
This press release drove me crazy the way it was written, both insulting and deceptive, but the research sounds very interesting. They mention doing arithmetic, which would make it pretty general. I'd love to know if they have any means of doing loops. There's mention of Turing machines in the press release but I'd guess that's just something that popped up in the weird garbage can of the authors mind while grunting the thing out.
Something like this combined with the ideas of 'amorphous computing' (think tons and tons of tiny unreliable processors communicating over short-range links, computation happening primarily as a large-scale emergent property of the whole system) might help things get truly weird in the future... thinking RNA paint? Sure, OK.
I'm not an RNA guy, but as far as I can tell, the highlight from the paper is that they've shown that you can computationally design these networks in an RNA-based system.
This kind of thing has been done with genetic circuits (DNA): http://science.sciencemag.org/content/352/6281/aac7341
I guess the somewhat interesting question is, 20+ years from now, will biological computers be DNA or RNA based?
The entire press release is cringeworthy.
What a world! There is far more in your sentence than the average geek might understand if she reads it too quickly. It could be nice to help everyone understand while keeping the humouristic touch.
(I only glanced through the OP, though.)
Reaction to article: WTF is this garbage? We've been able to build RNA sequences forever. Anyone who has read ANYTHING about bioinformatics knows RNA works like 'computer code' to tell a cell how to make a protein. This is high school level biology.
'The researchers made their invention by first modelling all possible RNA sequence interactions on a computer.' No, they didn't. Jesus Christ no they didn't. I don't know what they actually DID, or what they even CLAIM to have done, but they did not do this. First off, there is no limit to the number of possible RNA sequences. Secondly, there is no limit to the variety of possible interactions. You know Folding@Home? It simulates protein folding? That's what happens after you feed some RNA into a cells protein factory. So even if you figured out the infinite variety of proteins possibly created by the infinite variety of RNA sequences interacting with themselves and all the rest of the cellular contents (including potential viruses or bacteria that snuck in), then you'd have to solve every single folding problem on all of them.
Once you get to the actual words of the actual people doing the science, hey magic, it actually makes sense and sounds cool! But WOW, whoever wrote that press release up to the point where the closing quotation starts? That person needs to never write a press release for a scientific finding ever again. Humanity will be better off!