If I understand correctly, they demonstrated gates in living cells. Which immediately raises further questions, can they also build circuits, where do I get that stuff and how does IO work?
Yes, they can build circuits. See the supplementary materials, which also contain all the how-tos (ProTip: although most journal articles are behind paywalls, the materials & methods stuff is often freely accessible, at least for the Nature and Science journal families. It's also often a lot easier to read and understand for the lay person who isn't already familiar with the field in questions, assuming that your interest is more in the engineering than the science end - I see many papers whose theoretical validity I am incompetent to assess, but whose experimental process I am able to understand.) http://www.nature.com/nnano/journal/vaop/ncurrent/extref/nna...
EDIT: I spoke too soon. Section 6 of the supplementary material, which explores the idea of building more complex circuits and the scaling issues therewith, is unfortunately missing right now. I notified the authors and I'm sure that will be fixed soon. Meanwhile, here's what they say in the paper:
The architectures described here are capable of processing two input bits at most. However, the outputs from two 'processors’ can be relayed to a third and increase the processing capacity (limited by the possible number of unique gate-key systems that can be designed). Importantly, the scaling of our design is linear rather than logarithmic, with errors in robot activation propagating in the order of the root sum square of the number of unique robot types in a system (Supplementary Note 6). Figure 4 presents a schematic of a hypothetical 4-input bit architecture for an imaginary task of controlling three therapeutic molecules simultaneously, where only the number of robots comprising the system limits its capacity. The basic concept we describe can be scaled plausibly to exceed the capacity of older 8-bit computers such as a Commodore 64 or Atari 800, which many of us had experience of as children.
(isn't it nice that 'webDNA actually means DNA programming on the web' instead of some marketing department BS?) The raw material DNA strands are available from commercial suppliers.
IO is currently done by fluorospectroscopy, which is very common in experimental biology - you make compounds that flouresce inside the cell, then pull out some blood/do a biopsy/chop up the host if it's something very primitive and measure the degree of flourescence. So your unit testing is basically probabilistic rather than deterministic.
> Where you get that stuff: the robots are designed with CADnano, which is free software: http://cadnano.org/
I'm sorry, but there's CAD software for designing DNA structures now?! When did this happen? As someone tragically unaware of recent advances in the biosciences, I'd love to know more about this. I see the structures in screenshots of their CAD software, and then micrographs of the resulting real DNA(?) but how do you go from A to B?
Edit: Actually, for anyone reading this: what resources would you recommend for getting up to speed on cell biology, modelling of DNA and the like for someone with a solid science background? (I ask as someone who's done a lot of computer modelling, but more from a solid state physics point of view) I'd just like to be able to read articles like this with a slightly more critical eye...
by comparison, a transistor currently reliably scales down to 10 nm and the theoretical limit is probably close to 2-5 nm, and can be flipped at a rate in the range of Gigahertz. Being a designed system, the 100 nm scale is very close to the absolute smallest you can design a DNA logic gate, and I guarantee you it will never flip as fast as a electronic. And then just handling DNA is not trivial in general.
>DNA computing is fundamentally similar to parallel computing in that it takes advantage of the many different molecules of DNA to try many different possibilities at once.[11] For certain specialized problems, DNA computers are faster and smaller than any other computer built so far. Furthermore, particular mathematical computations have been demonstrated to work on a DNA computer. As an example, DNA molecules have been utilized to tackle the assignment problem.[12] Aran Nayebi[13] has provided a general implementation of Strassen's matrix multiplication algorithm on a DNA computer, although there are problems with scaling. In addition, Caltech researchers have created a circuit made from 130 unique DNA strands, which is able to calculate the square root of numbers up to 15.
Perhaps it can be used in nanobots. Also being slow isn't necessarily a disadvantage if it could be made much cheaper. Biology can rapidly self-replicate out of everyday materials. It can also run efficiently on chemical energy and even sunlight. Though these likely don't apply to this research, just biocomputers in general.
> Also being slow isn't necessarily a disadvantage if it could be made much cheaper.
silicon (processed sand) will always be cheaper. In order to make (nano) artifical biological parts, you have to use really expensive chemical building blocks, or, if it's biologically (and not petrochemically) derived go through expensive finishing steps.
> Biology can rapidly self-replicate out of everyday materials.
Then you're adding at least a micron of space overhead.
> It can also run efficiently on chemical energy and even sunlight
Also true of mechanical parts, for example palladium nanoparticles running off of peroxide, or light-driven silver nanoparticles. running off of sunlight is a nontrivial process for biology (turns out oxygen wrecks a lot of really important enzymes), and biology is actually not very good at it.
I wasn't sure which article to submit, since this was on gimodo, popular science, and various other places, but this (albeit behind a paywall) is the original journal article.
Searching on the title and appending 'filetype:pdf' for Google will often return a preprint or a copy from the authors' institutional website.
As for the content - wow, amazing work. I'm no biotechnologist but this seems like the gateway to an algorithmic approach to medicine. I wonder if it could be turned around to measure intra-cellular processes and send out signals by producing nonreactive chemical tracers...oh yes, it's addressed on page 4.
How long before we see biological botnets? Fascinating stuff.
And to quote Gibson: the future is weird.