Invited: Synthetic Neuromorphic Computing in Living Cells

While biological systems are inherently fuzzy and contain imprecise parts that collectively interact, synthetic computation in living cells is mostly inspired by precise computer engineering principles. Here, we demonstrate that neuromorphic synthetic genetic circuits can be engineered in living cells, while monolithically integrating signal-processing and decision-making. Such circuits exhibit perceptual behavior of artificial neural networks, to build intelligent systems with collective computational capabilities. First, we transferred principles and architectures of artificial neural networks to synthetic gene networks. Then, we applied them in Escherichia coli cells to create fuzzy logic (e.g. smooth functions) exploiting cooperativity, feedback loops and binding reactions. In addition, our circuits can be controlled according to the gradient-descent rule and back-propagation algorithm. We expect that morphing rules shared across bio-inspired systems, will pave the way for emerging industrial and therapeutic applications with adaptive engineered cells.