An Engineered Multicellular Bacterial Network for Simultaneously Answering Multiple Computational Decision Problems.

Living cell-based computers are in their infancy and answering multiple computational decision problems by a single system remains a key challenge. Here, we demonstrate an artificial neural network type architecture implemented with molecular-genetically engineered bacteria that answer four computational decision problems by identifying four types of prime numbers, including cluster prime, Euclid prime, safe prime, and Lucas prime, within the range of 0-9 in a chemical space. First, we demonstrated that the network consisting of four engineered cells classified two prime number families, namely cluster and Lucas prime numbers. Next, we scaled up the four-cell network to a six-cell network by introducing two new engineered cells and demonstrated that the new network classified four prime number families. Questions were asked to the bacteria by applying chemicals in binary patterns, and the answers were obtained from the distinct expression patterns of multiple fluorescent proteins. Each bacterium was engineered with synthetic gene regulatory networks such that the system chemistry followed the mathematical nature of an artificial neuro-synapse module. Collectively, the molecular-genetically engineered bacterial population formed a single-layered artificial neural network type architecture in liquid culture to perform the overall computation. The work may have implications in synthetic biology, biocomputing, and biologically implemented AI wetware.