PLACE: , Wyss Insitute for Biologically Inspired Engineering, Harvard University, Boston, USA
DATE: August 17-21, 2015
Full paper (track a) and extended abstract (track b) submission: Mar 22 2015
Notification of paper acceptance: May 8 2015
Revised manuscripts due: May 28 2015
Poster (track c) submission: May 22 2015
Notification of poster acceptance: June 12 2015
Early registration deadline: July 3 2015
See the conference webpage (http://dna21.wyss.harvard.edu/) for more information.
Highlights from DNA21
First session on August 17, 2015:
Chairman: Andrew Philips
Boemo et al. presented a method for mapping a logical formula, expressed in a propositional calculus, to a localized DNA strand displacement circuit. Each propositional variable is represented by a track and a DNA walker, and the geometry of the track is computed automatically for a given error tolerance. Probabilistic model checking is used to optimise the circuit design. This work is a step towards the automated design and analysis of localised DNA strand displacement systems.
Lakin and Stefanovic presented the design and analysis of an adaptive DNA strand displacement circuit for supervised learning. The circuit supports long-running computation by replenishing circuit components from an inactive buffer. Simulation of the circuit demonstrated learning of linear functions from observations, by stochastic gradient descent. Implementing long-running strand displacement circuits such as these represents an important challenge in the field of DNA computing.
Second session on August 17, 2015:
Chairman: Yonggang Ke
Shelley Wickham from Harvard University presented high-quality assembly of a group of “O barrels” with diameters up to 90 nm. Shelley showed the “O barrels” can served as a versatile platform for a wide range of applications, such as 3D molecular pegboard, nanoantenna, controlled cellular delivery.
Inspired by super-resolution DNA-PAINT, Alexander Johnson-Buck demonstrated a amplification-free, high-specificity (500-fold discrimination) method (SiMREPS — Single Molecule Recognition by Equilibrium Poisson Sampling) for probing microRNA at single-molecule level with almost zero background noise.
Sherry Wang presented simulation approach for designing ultraspecific nucleic acid hybridization. The experimental results showed between a 200- and 3000-fold specificity, a dramatic increase in comparison to 25-fold specificity for conventional approaches.
Second session on August 18, 2015:
Chairman: Jinglin Fu
Prof. Yannick Rondelez from Tokyo University presented a DNA-based reaction networks that controlled the degradation of individual DNA species. He developed a thermal dynamic model to predict the reaction networks which was in good agreement with the experimental results.
Dr. Guillaume Gines from Tokyo University presented a DNA circuits on microscopic beads. Using fluorescence imaging, it was possible to monitor the activation and diffusion of the DNA circuits reaction networks.
Mr. Nicholas Schiefer from Erik Winfree Lab at California Institute of Technology presented a universal strategy of designing the tile assembly model based on chemical reaction networks. His model will be useful to control the selfassembled structures by molecular inputs. Mr.Schiefer currently is an undergraduate student. His talk has showed that DNA computation and nanotechnology has attracted a broad interest of STEM students.
Third session on August 18, 2015:
Chairman: Elton Graugnard
Dr. Rothemund gave an inspiring presentation highlighting control over integrating DNA origami with semiconductor technology, combining the most powerful molecule nature has made with the most powerful technology humans have made. As an exquisite example of such control, he and postdoctoral researcher Dr. Ashwin Gopinath demonstrated the precise placement of DNA origami within photonic crystal cavities. With applications in on-chip single-photon light sources, the 94% yield for precision origami placement was key to achieving controlled coupling of origami-bound fluorophores to the resonant modes of the photonic crystal cavity. These results demonstrate the potential for DNA nanostructures to solve several current challenges in optical computing.
Toma “Tommy” Tomov presented his group’s progress on using single-molecule FRET to study computer-controlled DNA-based molecular machines. Using a novel hairpin fuel approach to speed the steps of a bipedal DNA motor, Tommy achieved an 88% yield for six steps along a DNA origami track. Using microfluidics to pulse and purge fuel strands, he demonstrated a yield of 44% for 32 steps along a track across two origami. These results are exciting “steps” toward programmable long-distance travel of molecular motors.
First session on August 19, 2015:
Chairman: Shinnosuke Seki
Lila Kari: Any DNA sequence can be converted into its Chaos Game Representation efficiently. CGR is a 2D-image. Given a set of DNA sequences, by computing an image distance between any pair of CGRs of DNA sequences in the set and visualizing the distances using multidimensional scaling, we can obtain the Molecular Distance Map of the set. The MDM can be used to compare the similarities between sequences in the set. By this method, they analyzed a dataset of 3176 complete mitochondrial genomes and found that the sequence most different from the anatomically modern human is cucumber.
Trent A. Rogers: In this talk, an extension of the abstract tile assembly model (aTAM) was proposed, in which a tile is allowed to be flipped horizontally and vertically. The authors first investigate the computational power of the extended model called the reflected tile assembly model (RTAM). The RTAM turns out to be not Turing complete at temperature 1, where cooperative binding is not available, while it is Turing complete at temperature 2, as in the case of aTAM. Then they examine the tile complexity of the n by n square at temperature 1. For even n, it is infinite, that is, it is impossible for an RTAM system to self-assembly the n by n square in this case at temperature 1. In contrast, for odd n, it is n, whereas in aTAM, it is conjectured that 2n-1 tile types are necessary. A characterization of finite shape self-assemblable by a mismatch-free RTAM systems is also given.
Scott M. Summers: Cook, Fu, and Schweller conjectured in SODA 2011 paper that the temperature-1 tile complexity of an n by n square in 3D is O(log n/ log log n). This paper gives an affirmative answer to this conjecture. The proposed construction uses only two layers z = 0, 1. It is due to the 3D temperature-1 general encoding scheme, similar to the 2D temperature-2 optimal encoding scheme by Soloveichik and Winfree.
Second session on August 19, 2015:
Chairman: Matthew Patitz
Natasha Jonoska from the University of South Florida gave a presentation titled “Molecular Ping-Pong Game of Life on a 2D Origami Array” based on work with Ned Seeman of New York University. While previous theoretical and experimental work has been done on the simulation of cellular automata, it has typically been focused on 1D cellular automata because of the static nature of the self-assembling components being utilized. However, Jonoska and Seeman propose a method of creating dynamic components capable of simulating 2D cellular automata in which DNA origami tiles are functionalized with differing types of fluorescent dyes. This would potentially allow an experimentalist to shine light of different wavelengths, calibrated to the different dyes, resulting in localized heating around the dye molecules which could cause particular DNA duplexes to melt and targeted tiles to detach and/or prevent their attachment. This mechanism, along with clever design of tile arrays, could provide a sort of “global clock” and synchronization of distributed computations via environmental control, and enable a wide class of computations to be performed via self-assembling tiles.
Cameron Chalk presented work by himself and Bin Fu, Alejandro Huerta, Mario Maldonado, Eric Martinez, Robert Schweller, and Tim Wylie, all of the University of Texas Rio Grande Valley, from a paper titled “Flipping Tiles: Concentration Independent Coin Flips in Tile Self-Assembly”. He showed how tile-based self-assembling systems which incorporate nondeterministic tile attachments which would typically be controlled according to the concentrations of the competing tile types can instead be simulated by systems in which the nondeterminism can be fairly split between alternatives, regardless of the relative concentrations of tile types, thus removing the need to carefully balance and control such concentrations. Furthermore, these simulating systems only require a constant increase in scale factor over the original systems.
Shinnosuke Seki gave a presentation on the paper “Efficient Universal Computation by Molecular Co-Transcriptional Folding”, which includes his work with Cody Geary, Pierre-Etienne Meunier, and Nicolas Schabanel. He presented Oritatami, their abstract mathematical model of systems composed of molecules which fold following prescribed rules during “transcription” so that the components of a growing line fold to form bonds with each other. He showed how it is possible to specify a single (infinite, repeating) sequence of components to be transcribed so that the growing and folding structure is capable of universal computation, with the output encoded into the shape of the resulting complex.
Third session on August 19, 2015:
Chairman: Yossi Weizmann
The work entitled “Designs and algorithms for DNA Folding of Custom 3D Polyhedra,” presented by first year PhD student, Abdulmelik Mohammed, introduces an algorithmic approach and software pipeline for the generation of 3D polyhedral DNA origami structures. Focusing on non-crossing routings, the strategy is based on three major requirements: (1) the DNA mesh should be triangulated to support structural rigidity; (2) each strand’s edge should be settled with one double helix to provide effective usage of DNA base-pairs; (3) the scaffold strand design should prevent self-crossing to avoid strands overlap and entanglement. To meet these requirements a triangulated polygonal model has been designed by the 3D software, in which the scaffolds have been routed according to an A-trail type in a Eulerian way. The method offers an efficient software pipeline, allowing for the synthesis of large and complex polyhedral DNA structures. The successful design has been supported by Transmission Electron Microscopy and Cryo-Electron Tomography images.
Fourth session on August 19, 2015:
Chairman: Björn Högberg
The presentation by Arne Schmidt concerned theory of algorithmic tile assembly. Arne Schmidt presented an overview of tile assembly history. He proceeded to show how to decompose shapes into rectangles to form tiles for various 2D shapes at different temperatures; without holes and with holes, and using recursion on all tiles.
Thim Strothmann indtroudced the field of programmable matter, in particular in the context of using modular and swarm robots. He compared it to DNA self-assembly, wound healing and development. Thim introduced their theoretical model called ameobot models, in which particles without a common compass or communication capability move, and extend connections between each other, on a hexagonal grid. The particles can “elect” a global leader particle using only local neighbor communication. Once you have a leader, shapes can be built in linear time using that particle as a seed.
Petr Sulc talked about his interest in RNA nanotechnology and how it can easily be interfaced with biological processes. He presented his model of RNA that can be efficiently simulated on GPU clusters. The model reproduces the RNA helices without any explicit solvent (helps to speed up simulation) using model interactions that are simplified models of the real physical interactions between RNA bases. The model is fitted to reproduce the melting temperatures as predicted by the Turner’s nearest neighbor model for RNA. Petr presented several cases where their model simulation data fits experimental data for equivalent systems: RNA hairpin unzipping, RNA strand displacement (here he compared with DNA experimental data), RNA plectonemes.
First session on August 21, 2015:
Chairman: Kurt Gothelf
In the morning session of the Nanoday Friday, Ned Seeman from New York University gave a talk entitled “DNA and Nanotechnology” where he among other things presented new aspects of crystallization of 3D DNA arrays and how the color of the crystals can be controlled by application of dye labeled strands. It was shown how DNA hybridization can proceed within the crystals after their formation by soaking the crystals in a solution containing a labeled strand. Seeman also talked about DNA robotics and self-replicating systems.
In the second talk Hao Yan from Arizona State University presented new ways to fold DNA origami including new complex wireframe DNA origami nanostructures in 2D and 3D. A design principle for forming single stranded DNA and single stranded RNA origami was also introduced. A project made in collaboration with Ned Seeman on new tile structures for formation of DNA crystals and structural data obtained by X-ray crystallography was also introduced.
In the final talk of the morning session Mark Bathe from MIT presented a general computational method to convert a given model into a DNA origami structure. In one example this method was applied to the design of custom J-aggregate chromophore assemblies for mimicking energy transfer in nature.
Second session on August 21, 2015:
Chairman: Hao Yan
In the second session of the Nanoday, Kurt Gothelf from Aarhus University first presented elegant work of using DNA nanoscaffold to precisely control the positioning of conducting polymers with defined patterns. He also showed a new strategy to achieve regioselective and site specific DNA protein conjugation.
In the second talk, Yuanchen Dong spoke on behalf of Dongsheng Liu from Tsinghua University about idea and experimental implementation of frame guided assembly where they used both nano particles and DNA origami nanostructures as frame to guide the geometrical assembly of lipid layers to to emulate Nature’s concept of scaffolded assembly of cellular membrane architectures.
This session with a nice talk from Lulu Qian from Caltech, who presented a novel method to create maze like 2D patterns using truchet tiles using combinatorial self-assembly from a deliberately designed DNA origami nanostructure.
Third session on August 21, 2015:
Chairman: Fritz Simmel
The first speaker, David Zhang, gave an impressive talk, in which he demonstrated how DNA circuits based on DNA strand displacement reactions could be made extremely sensitive to single nucleotide changes in sequence. This principle can be used for DNA mismatch detection, which is the basis of one of the first commercial applications for DNA circuits in molecular diagnosis.
In the following talk, Andrew Ellington advertised the utilization of DNA functionalized surfaces for DNA computing and molecular programming. He argued that the field could benefit immensely from fabrication methods and screening strategies initially developed for next generation sequencing. He demonstrated the use of gene chips as a platform for DNA circuits, and also showed how a DNA walker could traverse the surface of DNA functionalized beads.
As the final speaker, James Collins provided a historical perspective on the development of synthetic gene switches in the early days of synthetic biology. He then showed several applications of such switches in the context of biomedical diagnosis such as their use as in vivo memories recording the molecular history experienced by them. Another highlight of the talk was the demonstration of Ebola diagnostic sensors based on toehold switches, which could be operated in a paper-based format using cell-free gene expression.
Last session on August 21, 2015:
Chairman: Lulu Qian
Fritz Simmel described research progress on components for future “artificial cells”: how to use transcriptional circuits (also called genelet circuits) to control clocked production of functional RNA molecules such as RNA aptamers (Franco et. al., PNAS 2011), how to use water-in-oil droplets to compartmentalize biochemical circuits and study the stochastic behavior of these circuits in a small volume comparable to a cell (Weitz et.al., Nature Chemistry 2014), how to create artificial membrane channels using self-assembled DNA nanostructures for applications such as single molecule sensing (Langecker et. al., Science 2012), and how to introduce communication among compartmentalized genetic circuits using arrays of water-in-oil droplets and diffusible inducer molecules, including the demonstration of an AND gate with input molecules from two types of neighboring droplets (Weitz et. al., JACS 2014).
Ed Boyden talked about a groundbreaking new imaging technique called expansion microscopy that enables super-resolution imaging of biological samples including cultured cells and brain tissue using traditional diffraction-limited microscopes (Chen et. al., Science 2015). This technique transfers the structural information within a biological sample into a swellable material and physically expands that imprinted material 4.5 times evenly in all three dimensions, such that the structure of interest is preserved and magnified. The target biomolecules were labeled by an antibody linked to a DNA oligonucleotide that is complementary to another oligonucleotide modified with a chemical dye on one end and a methacryloyl group on the other. After the monomer of the swellable material was injected to the sample and polymerization took place, the methacryloyl group was anchored into the polymer network. All endogenous structure in the sample was then digested so it will not resist the expansion, and the sample was placed in water and the polymer network spontaneously expanded. After expansion, the sample was mostly water thus was transparent. With this technique, scattering is significantly eliminated and a much larger volume of biological sample can be successfully imaged, compared to the super-resolution imaging techniques. So far this technique has been applied to image microtubules in cells, mouse brain tissue, and zebrafish embryo with 60 to 70 nanometer resolution. Larger expansion and higher resolution could potentially be achieved through methods such as tuning the properties of the swellable material.
Wesley Wong talked about quantitative methods for biophysical analysis of molecular interactions. The centrifugal force microscope (Halvorsen and Wong, Biophysical Journal 2010) was introduced as a high-throughput alternative to optical tweezers for measuring force-extension curves. A second-generation design, constructed using a 3D printer, is now available for use in commercial centrifuges (Hoang et. al., Biophysical Journal 2015). In order to apply the technique to easily measure a variety of molecular interactions, they developed a “molecular instrument”: a DNA nanoswitch based on “1D DNA origami”, to which the molecules of interest can be attached, that exhibits a large change in length upon disruption of the bond (Halvorsen et. al., Nanotechnology 2011). The DNA nanoswitch can also be used for low-cost, quantitative analysis of molecular interactions in a simple gel assay (Koussa et. al., Nature Methods 2015).