The quest for energy-efficient computation inspires researchers to explore novel materials and methods, and a team led by Tim E. Veenstra, René van Roij, and Marjolein Dijkstra from Utrecht University now demonstrates a promising pathway using self-assembling DNA tiles. They reveal that these tiny structures, programmed to build multiple configurations, perform basic computational tasks akin to simple machines when driven by controlled interactions. The researchers establish design principles for these dynamic assemblies, showing they can count, perform mathematical operations, and even recognise specific patterns, all while consuming minimal energy. This work integrates sensing, memory, and action within a single physical system, offering a blueprint for future computation embedded directly into materials like DNA, proteins, and colloids.

Processing opens new avenues for computation embodied in matter. Researchers demonstrate 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. These systems achieve this functionality when equipped with programmable interactions that drive controlled transitions between these structures. By establishing design rules for multifarious self-assembly while carefully managing the energy input required to drive these transitions, the team demonstrates that these systems can execute a wide variety of tasks, including counting, calculating modulo functions, and recognizing specific input patterns.

DNA Nanostructures Implement Finite State Automata

Scientists have created a physical system that mimics the behaviour of a finite state automaton, but instead of using electronics, they employ self-assembling DNA nanostructures. This approach uses distinct structures to represent different states, and controlled transitions between these structures to process information. A key feature of this system is its ability to reuse building blocks, or particles, across multiple structures, simplifying the design and reducing complexity. The research addresses several challenges inherent in translating abstract computational concepts into a physical system.

One challenge arises when a single input could trigger opposing transitions, potentially destabilizing the system. To overcome this, scientists duplicated states, effectively creating parallel pathways for transitions and avoiding conflicting actions. Another challenge involves odd-numbered cycles of states, which require alternating particle libraries during transitions. Again, duplicating states breaks the cycle into an even length, ensuring consistent particle usage. Maintaining the system’s ability to reuse particles, or ‘multifariousness’, is crucial for scalability and practical implementation, and is achieved through careful design of the structures and transitions.

The design principles emphasize the importance of reusing particles, adhering to constraints imposed by the physical system, and utilizing non-reciprocal interactions to ensure a directional flow of information. An algorithm automatically generates target structures that meet these design constraints, maximizing particle reuse and ensuring reliable operation. The system operates by receiving an input, disassembling the current structure, and simultaneously assembling a new structure using the released particles, effectively transitioning to a new state. This process repeats as the system receives new inputs and processes information. This research is significant because it explores a new paradigm for computation, moving beyond traditional electronics to harness the power of physical self-assembly. This could lead to ultra-low power computing, biocompatible systems using biological materials like DNA, and novel architectures that are difficult to achieve with conventional manufacturing techniques.

DNA Tiles Compute Binary Counting Sequences

Scientists have achieved a breakthrough in physical computation by constructing a system of DNA tiles capable of performing tasks analogous to those of finite-state automata. This work demonstrates how assemblies of DNA tiles, designed to self-organize into multiple target structures, can execute computational operations through controlled dynamical transitions between these structures. The research establishes design rules for creating these multifarious self-assembling systems while carefully managing the energy input required to drive the transitions. Experiments revealed that the system successfully counts binary inputs, accurately determining the number of “1” bits in sequences up to three digits long.

By designing a system with four distinct target structures representing states 0 through 3, scientists demonstrated modulo 4 computation with a fidelity of at least 95% across 21 individual simulations for each of the eight possible three-bit inputs. Errors were limited to single missed transitions, indicating a robust and reliable computational process. Further extending this capability, the team designed a Brownian finite-state automaton to compute i mod 3 for binary inputs. This more complex computation required careful consideration of potential conflicting transitions, where the same input could trigger opposing state changes.

By leveraging the multifarious nature of the system, effectively doubling the number of target structures while reusing particle libraries, scientists overcame this challenge. Simulations with 15 different four-bit input sequences demonstrated that the automaton correctly computed the modulo 3 result in at least 95% of the trials, confirming its ability to track bit order and perform more sophisticated calculations. The system utilizes shared libraries containing between 284 and 295 species, minimizing the overall complexity while maximizing computational power. These results establish a foundation for designing more complex computational automata embedded within materials ranging from DNA and enzymes to proteins and colloids.

DNA Tiles Compute and Recognize Patterns

This work demonstrates a physical computing system based on the self-assembly of DNA tiles, capable of performing computations analogous to those of finite-state automata. Researchers successfully designed assemblies that, when driven by controlled transitions between different structures, can execute tasks including counting, calculating modulo functions, and recognizing specific input patterns. The system achieves this through a combination of multifarious self-assembly and the careful management of non-reciprocal transitions, effectively integrating memory, sensing, and actuation within a single physical platform. Notably, the system reliably distinguishes between inputs containing a specific bit pattern and those lacking it, achieving high probability for correct identification.

A key challenge addressed by the team was preventing premature transitions that could destabilize the system, and they overcame this by implementing a “non-reciprocity budget” and design rules that ensure each input triggers only a single transition. While the current demonstration focuses on specific tasks, the underlying principles are broadly applicable to any system with a finite number of states and controlled transition mechanisms, opening avenues for the creation of adaptive and energy-efficient materials. Researchers acknowledge that controlling these transitions remains a critical aspect for future development, and further work will likely focus on expanding the complexity of computations and exploring diverse material platforms for implementation.

👉 More information
🗞 Counting, Computing, and Pattern Recognition with Self-Assembling Non-Reciprocal DNA Tiles
🧠 ArXiv: https://arxiv.org/abs/2510.19503