06. Jeremie Marlhens: Designing Multistable Systems with Biomolecular Hopfield Networks

Abstract. Multistate systems are fundamental to many biological functions, particularly in development and differentiation. Although engineering simple bistable systems is possible, designing arbitrary multistate systems systematically remains a challenging task. We address this by proposing a molecular Hopfield network, a synthetic molecular reaction network that exhibits multi-stability. Unlike existing approaches that require optimization, our network is derived directly using the Hopfield network Hebbian rule, enabling instant design of multistable systems of any size.
Our model utilizes a synthetic molecular reaction network based on two key mechanisms: protein sequestration and positive feedback. These mechanisms confer bistability to each node of the network, composed of two proteins. Pairs of proteins form complexes through sequestration, representing positive and negative values. As the sequestration rate approaches infinity, the system converges to the continuous Hopfield network, inheriting its known properties, such as capacity and robust stability. This convergence provides a theoretical foundation for controlling multiple stable states in biological systems.
The multistability in our simple structure reflects the behavior of developmental and cancer networks, suggesting potential applications in synthetic biology. By mimicking active genetic networks during development, our model provides a novel method for engineering controllable multistate biological systems. This research paves the way for creating stable synthetic networks with predictable behaviors.
Bio. Jérémie Marlhens is a PhD student at TU Darmstadt, Germany, specializing in cell-free systems. His research centers on designing biological circuits through mechanistic modeling and learning from data, aiming to understand and control complex biological behaviors. With a background in biology and applied mathematics from École normale supérieure de Lyon and École Centrale de Lyon, Jérémie's work integrates high-throughput methods and machine learning to tackle challenges in synthetic biology. His projects include synthetic biology, high-throughput RNA analysis, and biosensor design, underscoring his commitment to interdisciplinary approaches in circuit design and biological system control