3
$\begingroup$

I'm a computer science guy, recently crossing over to do some research in computational biology on RNA secondary structure prediction. While looking through the materials I got a crazy idea, what if you could design a synthetic ribosome that acts as a Turing Machine on strings of RNA? (you could do similar things with DNA enzymes)

The immediate applications could be stuff like correcting mutations in RNA to cure genetic diseases, which is a relatively simple computation:

  1. Check if a string is bad.
  2. If so, fix bad letter.

Another could be "shredding" DNA/RNA of viruses:

  1. Check if a string is a virus.
  2. If so, shred it.

I'm not sure the status of the literature on this is, or synthetic biology's ability to create something like that. Do you think it is a good idea to pursue this idea or not?

I realize this is a soft question. Perhaps a better one would be, is this possible?

$\endgroup$
  • 1
    $\begingroup$ There are some papers on simple computations using nucleic acids, if nobody answers I'll look them up later. As for your examples, nature already does some of them (see RNA interference). And for correcting genetic diseases, the hard part is actually correcting the sequence in all cells in your body, not the detection itself. $\endgroup$ – Mad Scientist Jun 20 '14 at 18:56
  • 1
    $\begingroup$ Where would the 'table' of the machine be? In other words how would the 'bad string' be recognised? $\endgroup$ – Alan Boyd Jun 20 '14 at 18:57
  • $\begingroup$ @AlanBoyd An immediate thought would be to keep a copy of the bad string, say a mutated RNA strand, in the machine. Combine it with an enzyme bound to it with 2 functions, check if the letter it's currently bound to matches the current input letter, and slide right. That could do the check step. For the replace step you could do the same backwards and then splice the correct protein in. For a general Turing Machine, you could keep a string of the current state in RNA, keep the rules in another string, and slide along it to find the rule and execute it. $\endgroup$ – Mike Flynn Jun 20 '14 at 19:31
  • $\begingroup$ I'm no expert at this but I don't think you need a Turing machine to achieve this, at least for DNA, since in conventional sense (not including carcinogenic factors such as UV) the most probable stage in which a DNA mutation can happen is during DNA replication so DNA polymerase (en.wikipedia.org/wiki/DNA_polymerase) engages in proofread to ensure accurate DNA replication, and then the DNA can get back into being heterochromatic (eventually). This will be more difficult for RNA as they are single stranded although they do fold onto themselves. $\endgroup$ – Bez Jun 20 '14 at 19:40
  • $\begingroup$ Also there many many strands of RNA that are produced through alternative splicing, VDJ recombination and RAG imprecise rearrangements for antibody production, which can give rise to infinite possible RNA productions. So how would you know what is "good" and what is "bad" in terms of RNA and its production. Half the time we don't even know what they do and what their roles are and the only reason we call something as a "bad" or "faulty" RNA is because it is having a negative effect not because we understand the precise role of the sequence, which means a theoretical table would be useless. $\endgroup$ – Bez Jun 20 '14 at 19:59
2
$\begingroup$

I'm no expert at this but I don't think you need a Turing machine to achieve this, at least for DNA, since in conventional sense (not including carcinogenic factors such as UV) the most probable stage in which a DNA mutation can happen is during DNA replication so DNA polymerase engages in proofread to ensure accurate DNA replication, and then the DNA can get back into being heterochromatic (eventually). This will be more difficult for RNA as they are single stranded although they do fold onto themselves.

Also there many many strands of RNA that are produced through alternative splicing, VDJ recombination and RAG imprecise rearrangements for antibody production, which can give rise to infinite possible RNA productions. So how would you know what is "good" and what is "bad" in terms of RNA and its production. Half the time we don't even know what they do and what their roles are and the only reason we call something as a "bad" or "faulty" RNA is because it is having a negative effect not because we understand the precise role of the sequence, which means a theoretical table would be useless.

So in short, a turing style protein machinery is neither possible (due to almost infinite potential number of RNA produced) nor feasible (since the number of proteins that would have to check each RNA sequence would in turn have to be almost infinite). And none of these takes into account SNPs i.e. individual variations, which can produce proteins with different sequence but identical structure/function, since obviously any lethal variations would be selected against but thats a whole different argument.

$\endgroup$
  • $\begingroup$ It should be possible to create Wang tiles using DNA or RNA. Different colored edges could be represented by different sequences selected to bind to each other selectively (DNA and RNA are very good at this!). Wang tiles can be used to implement a 1D cellular automaton, which can implement a Turing machine -- it will keep a record of its complete history as a big wobbly mat. Wang tiles are collections of square tiles in which each edge has a certain color, adjacent edges having matching colors. They can produce infinitely extendable patterns that can't be periodic (similar to Penrose tiles). $\endgroup$ – Paul Harrison Jan 2 at 10:24
  • 1
    $\begingroup$ @PaulHarrison - Welcome to SE Biology. I really feel, like the original post over four years ago, you are on the wrong list. One reason there was only one answer to the original question is that few biologists know what a Turing machine is, and most of those who do feel it has no relevance with biology. I have even less idea what Wang tiles are. However if you feel that you have a useful point to make you can post your own question on SE and then answer it. I would suggest that you do that on SE Bioinformatics, though, as you are more likely to get a response there. $\endgroup$ – David Jan 2 at 14:40
  • $\begingroup$ @PaulHarrison - I don't see how your post targets the question in any way. It's more of a comment to the other answer I think. As David says, I indeed know what a Turing machine is but the tiles are over the top of my head too. I'm converting the answer to a comment of the other answer for now. Feel free to modify it to make it a more specific answer to this site in a biological context. Don't forget to include sources to your material. $\endgroup$ – AliceD Jan 4 at 13:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.