Yannick Rondelez explained his DNA toolbox. He uses enzymes; enzymes are rare at this conference. Polymerase and nickase can create a cycle of production of copies of a template. The copies can go on to prime other copying processes (including self-copying) and it can inhibit other copying processes. That’s a nice system for building reaction networks. That’s the toolbox. The computer takes target behavior, translates to a network, then to reactions, then to DNA.
What kinds of dynamic reaction systems can you make? You can make bistable, autoamplifying, cyclical, etc. What is really amazing is that you can evolve a reaction network in silico using a genetic algorithm! Then, you can make the stuff and show that a bizarre reaction network (like a square wave oscillator!). I gather he has not yet actually produced the square wave experimentally. It’s only 18 nucleotides, but hybridization/reaction rates need to be super-precisely defined, I expect. Then use a microfluidic droplet-based parameter space screening chip. That’s almost as good as an in vitro square wave oscillator.
Niles Pierce talked about RNAi. RNA interference uses Dicer enzyme to chop dsRNA into siRNA then to RISC, which retains the recognition domain to destroy other mRNA. Long stoy short: pick a gene and turn it off. What he is working on is transduction of mRNAx to activate siRNAy. This is a programmable way to turn off a specific gene in a specific location at a specific time. They introduce a strand displacement into the fertilized egg cell, then when a particular mRNA turns on in a specific daughter cell, it is transduced into siRNAY. That is both clever and might actually work.
Alphonso Rodriguez-Paton talks about probabilistic reactions with DNA toolbox. Probabalistic inference reminds me of the hidden Markov model I read about in “How to Create a Mind.” If you want to diagnose a disease, you have symptoms (signals) and you want to infer the reality (hidden variables). The more symptoms you have, the better. Detection, even of things that we don’t understand analytically, can feed into statistical analysis. But can we use the DNA to do these calculations? It looks like the DNA toolbox can probably do that. That’s neat: human readable test strip with some really fancy data analysis built right in.
Frits Dannenberg asked: what can be computed with DNA walker circuits? The Qian paper is the root of the work. Logic can be translated to DNA circuits/Ntworks… can these reactions be localized, sped up, leakage minimized? Will they compute at single molecule level? This is the same basic idea as reaction network computers, but uses molecular walkers as computers. Wickham walker uses a track with junctions. Complex, but reliable? Uses paths to compute. It was intersting, but I got distracted.
Adrien Pedirac talked about Spatial Programming: microfluidics were used to coordinate experimental conditions. He showed DNA and enzyme based waves and spirals. I wanted to do this when I got to Andy’s lab. Microfluidics allows for an environment in which predator-prey waves can propagate. Pedirac’s experiment generates amazing spirals and waves in 2D just like simulations predict (I even made some simulations myself). The reactors are DNA toolbox type. Microfluidics were used to control the environment that were based on PDMS and Quake valves.
Fei Zhang from Hao Yan’s group. Builds really small tiles. They have assembly rules but are not algorithmic. They have not made quasicrystals. Yet.
John Milligan – DNA circuits as real time detector of isothermal amplification. From the same lab as myself, ellingtonlab.org. He won best student presentation for a very nice theory and application presentation on using catalyzed hairpin reactions to detect real samples.