Heat-rechargeable computation in DNA logic circuits and neural networks - Nature

"Nucleic acid circuits using polymerases can maintain dynamic steady states by consuming nucleoside or deoxynucleoside triphosphates7,8, demonstrating diverse functions, such as multistate memories9 and nonlinear classification10. However, with a few exceptions11, they continuously consume energy even without input changes, leading to energy waste and limited operation times12. Enzyme-free nucleic acid circuits use only rationally designed components, making them more robust to environmental changes, such as temperature and salt conditions13, while performing complex information processing tasks, including Boolean logic14,15 and neural computation16,17."

"Yet, beyond equilibrium computation18,19, these systems cannot sustain information processing and respond to new inputs once the energy from unpaired bases or out-of-equilibrium concentrations is depleted. This limitation has hindered the development of enzyme-free circuits capable of advanced behaviours, such as iterative computation20 and learning21. Several strategies have been explored to achieve more sustainable computations in enzyme-free nucleic acid circuits, but the complexity remains limited to simple logic circuits with fewer than a dozen distinct molecular species18,22,23,24,25,26."

Nucleic acid circuits that use polymerases can maintain dynamic steady states by consuming triphosphates, enabling functions like multistate memories and nonlinear classification, but they continually consume energy even without input changes, causing energy waste and limited operation times. Enzyme-free nucleic acid circuits rely on rationally designed components and show robustness to temperature and salt variations while performing Boolean logic and neural computation. These systems cannot sustain information processing beyond equilibrium once available chemical energy is depleted, preventing responses to new inputs and limiting advanced behaviours like iterative computation and learning. Existing strategies including buffering and reversible reactions face scalability, autonomy, and waste-accumulation problems.

Read at Nature