Counting, Computing, and Pattern Recognition with Self-Assembling Non-Reciprocal DNA Tiles
Tim E. Veenstra, René van Roij, Marjolein Dijkstra · 2025
Read Counting, Computing, and Pattern Recognition with Self-Assembling Non-Reciprocal DNA Tiles by Tim E. Veenstra, René van Roij, Marjolein Dijkstra (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-19503.txt.
Introduction
Tim E. Veenstra, René van Roij, and Marjolein Dijkstra show that self-assembling DNA tiles can be made to compute, not just to form patterns. In their simulations, tiles fitted with programmable non-reciprocal interactions switch between target structures in a controlled way. That switching lets the assemblies act like simple finite-state machines.
The team sets design rules for assemblies that can form several different structures while keeping track of the energy needed to drive the changes. With those rules, the tile systems count, compute remainders, and recognize specific input patterns. Because memory, sensing, and action live in one material, the work points toward energy-efficient computing built into matter itself.
Catalog record
| CREATOR | Tim E. Veenstra, René van Roij, Marjolein Dijkstra |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 59 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-19503
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Counting, Computing, and Pattern Recognition with Self-Assembling Non-Reciprocal DNA Tiles Tim E. Veenstra, René van Roij, Marjolein Dijkstra Abstract Harnessing the intrinsic dynamics of physical systems for information processing opens new avenues for computation embodied in matter. Using simulations of a model system, we show that assemblies of DNA tiles capable of self-organizing into multiple target structures can perform basic computational tasks analogous to those of finite-state automata when equipped with programmable non-reciprocal interactions that drive controlled dynamical transitions between these structures. By establishing design rules for multifarious self-assembly while budgeting the energy input required to drive these non-equilibrium transitions, we demonstrate that these systems can execute a wide variety of tasks including counting, computing modulo functions, and recognizing specific input patterns. This framework integrates memory, sensing, and actuation within a single physical platform, paving the way toward energy-efficient physical computation embedded in materials ranging from DNA and enzymes to proteins and colloids. Biological systems display remarkable non-equilibrium behavior, continuously consuming and dissipating energy to sustain life. This energy drives diverse processes, from cargo transport by kinesin proteins, to the flagellar motion powered by proton transport, to ion pumps maintaining membrane potentials. Many biological materials feature building blocks that selectively bind to form complex, functional structures. Striking examples are protein condensates, dynamic molecular entities that assemble and disassemble to perform specialized cellular functions. Advances in bio-inspired nanotechnology have extended this principle to synthetic systems, using nucleic acids that interact via Watson-Crick-Franklin hybridization. Seeman and Sleiman 2017 Such interactions enable addressable structures, where each building block occupies a unique position within the target structure by binding selectively to designated neighbors. This principle has enabled the design and assembly of a wide range of structures using DNA tiles, Seeman 1982 ; Murugan et al. 2015 ; Osat and Golestanian 2022 ; Evans et al. 2024 DNA origami, Liu et al. 2016 ; Tian et al. 2020 ; Sigl et al. 2021 ; Liu et al. 2024a and DNA-grafted colloids. Biancaniello, Kim, and Crocker 2005 ; Nykypanchuk et al. 2008 ; McMullen et al. 2022 ; Rogers and Manoharan 2015 ; Rogers, Shih, and Manoharan 2016 ; Chakraborty et al. 2017 ; He et al. 2020 ; Moerman et al. 2023 These addressable systems have enabled applications in drug delivery, biological sensing, catalysis, and photonics. Rogers, Shih, and Manoharan 2016 Building blocks need not belong to a single structure; identical components can participate in multiple assemblies, as seen in proteins forming distinct complexes. Sartori and Leibler 2020 This multifariousness enables the self-assembly of several target structures from a shared library of building blocks. Murugan et al. 2015 ; Bupathy, Frenkel, and Sastry 2022 ; Metson, Osat, and Golestanian 2025a Each target structure corresponds to a minimum in a high-dimensional free-energy landscape, acting as attractors Keim et al. 2019 and making multifarious self-assembly analogous to associative memory in Hopfield networks. Hopfield 1982 Here we will see that even richer behavior can emerge when non-equilibrium dynamics is introduced. A particularly exciting class of non-equilibrium phenomena involves non-reciprocal interactions, where action-reaction symmetry is broken. Such interactions violate time-reversal symmetry and convert energy irreversibly into useful work, leading to rich dynamical behavior. Fruchart et al. 2021 ; Barrat et al. 2023 ; Dinelli et al. 2023 Non-reciprocity has been exploited to create predator-prey droplets, Meredith et al. 2020 ; Liu et al. 2024b active solids, Brandenbourger et al. 2019 ; Veenstra et al. 2025 motile particle clusters, Navas and Klapp 2024 and diverse patterns. Saha, Agudo-Canalejo, and Golestanian 2020 ; Rana and Golestanian 2024 ; Huang et al. 2024 Crucially, it enables systems to escape kinetic traps and cross energy barriers, Navas and Klapp 2024 ; Osat et al. 2024 ; Osat and Golestanian 2022 a property leveraged in multifarious assemblies. When building blocks exhibit non-reciprocal interactions, transitions between distinct target structures can occur in a directed and controlled manner. Osat and Golestanian 2022 By breaking detailed balance, non-reciprocity allows transitions between a sequence of energy minima, which effectively endows systems with “shape-shifting” behavior such that the next target structure nucleates and grows out of the previous one. Metson, Osat, and Golestanian 2025b Here, we leverage non-reciprocal interactions not only to drive transitions between multifarious target structures, but also to design information-processing devices that accept inputs, process them, and generate outputs. Specifically, we develop finite-state automata for basic computations in a Brownian system with well-defined and controllable transitions between discrete states. Our approach employs multifarious self-assembly, in which a shared library of building blocks can self-organize into multiple target structures, which we use as distinct states of a finite-state machine. Sipser 2012 ; Kwakernaak and Hecke 2023 ; Liu et al. 2024c Subsequently, we exploit non-reciprocal interactions to induce transitions between the different target states. We establish design rules to control these non-reciprocal transitions while budgeting the energy input (fuel) required to drive the non-equilibrium dynamics. Using this framework, we demonstrate systems capable of counting, computing modulo functions, and recognizing input patterns. Figure 1: (a) Schematic representation of two target structures, labeled S_{0} and S_{1} , both composed of a library that consists of nine different building block species. Particle colors in these sketches are arbitrary and only intended to distinguish different target structures, particles of the same color represent different particle species. Neighboring particles within the target structures bind with a strength \varepsilon as shown in the insets. (b) Schematic of a few of the intermediate steps during the transition S_{0}\rightarrow S_{1} as facilitated by the non-reciprocal interactions \lambda and the reciprocal interactions \eta between neighboring building blocks in the initial and the subsequent structure. Here we show that the particle of species \beta from target structure S_{1} replaces one of species \alpha from S_{0} , driven by the availability of four times the non-reciprocal energy contribution \lambda in the swap rate p_{swap} , as defined in Eq.( 4 ). The neighboring particles from two consecutive target structures (here S_{0} and S_{1} ) are bound together during the transition with bond strength \eta to stabilize the intermediate structures, as shown in the inset. Distinct states and Controlled Transitions We consider a multicomponent mixture of N_{s} DNA tile species on a three-dimensional cubic lattice, each with directional binding sites on its four faces in the xy -plane (Methods). These programmed directional interactions are designed to promote self-assembly into one of m distinct two-dimensional multifarious target structures S_{0},S_{1},\dots,S_{m-1} (Fig. 1 (a)). Each target structure comprises of N_{t}=18\times 16=288 tiles drawn from a shared library of species. The stability of these structures arises from directional bonds of energy -\varepsilon=-6.8k_{B}T between neighboring tiles. To enable controlled transitions between target structures, we introduce non-reciprocal “swap moves” between DNA tile species \alpha and \beta (Fig. 1 (b)). Osat and Golestanian 2022 When \alpha is stably bound in the low-energy state of structure S_{\ell} , swaps are rare. To overcome this, an energy input \lambda>0 —representing external fuel, enzymatic activity, or concentration gradients— is used to break an \alpha – \gamma bond in S_{\ell} and form a new \beta – \gamma bond. This occurs only when \beta\in S_{\ell+1} is adjacent to \gamma\in S_{\ell} (Methods). We found that stabilizing intermediate structures containing particles from both S_{\ell} and S_{\ell+1} is essential. Without this stabilization, S_{\ell+1}