Engineering a Microbiome Malathion Sensor Using Data-Driven Approaches

Malathion is a chemical compound commonly found in a wide variety of pesticides throughout the world. It is known to be toxic to birds and aquatic species, as well as slightly toxic to humans. In this work, we engineer a microbiome to act as bacterial sensors in various strains of bacteria to act as a Malathion detector. When Malathion is detected, the microbiome is engineered to fluoresce. Our methodology consists of first collecting time-series RNA sequencing measurements of a microbiome of Vibrio natrigens, Pseudomonas fluorescens, and Burkholderia Thailandensis as they are cultured in LB broth spiked with 0.615 mg/L of Malathion and also another batch cultured without any Malathion input. It is important to note that the growth curves of these strains growing in LB broth with and without Malathion were identical. From the gene expression measurements, we then identify the locations of the bacterial genomes which respond strongly to the Malathion chemical input using data-driven, Koopman operator approaches. Although the gene expression dynamics are highly nonlinear, our Koopman operator approach allows us to lift the dynamics to a higher-dimensional space where the dynamics are linear. In this new observable space, we make use of the observability gramian to identify the most observables modes of the system.