Recurrent neural chemical reaction networks that approximate arbitrary dynamics
Author(s)
Dack, Alex
Qureshi, Benjamin
Ouldridge, Thomas
Plesa, Tomislav
Type
Journal Article
Abstract
Many important phenomena in biochemistry and biology exploit dynamical features such as multi-stability, oscillations, and chaos. The construction of novel chemical systems with such rich dynamics is a challenging problem central to the fields of synthetic biology and molecular nanotechnology. In this paper, we address this problem by putting forward a molecular version of a recurrent artificial neural network, which we call a “recurrent neural chemical reaction network” (RNCRN). The RNCRN uses a modular architecture—a network of chemical neurons—to approximate arbitrary dynamics. We first prove that, with sufficiently many chemical neurons and suitably fast reactions, the RNCRN can be systematically trained to achieve any dynamics. RNCRNs with a relatively small number of chemical neurons and a moderate range of reaction rates are then trained to display a variety of biologically important dynamical features. We also demonstrate that such RNCRNs are experimentally implementable with DNA-strand-displacement technologies. A record of this paper’s transparent peer review process is included in the supplemental information.
Date Issued
2026-04-22
Date Acceptance
2026-03-11
Citation
Cell Systems, 2026
ISSN
2405-4712
Publisher
Elsevier (Cell Press)
Journal / Book Title
Cell Systems
Copyright Statement
© 2026 The Authors. Published by Elsevier Inc. 1 This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
License URL
Identifier
10.1016/j.cels.2026.101572
Publication Status
Published online
Article Number
101572
Date Publish Online
2026-04-22