Automated design of synthetic microbial communities

Abstract

In naturally occurring microbial systems, species rarely exist in isolation. There is strong ecological evidence for a positive relationship between species diversity and the functional output of communities. The pervasiveness of these communities in nature highlights that there may be advantages for engineered strains to exist in cocultures as well. Building synthetic microbial communities allows us to create distributed systems that mitigates issues often found in engineering a monoculture, especially when functional complexity is increasing. Here, we demonstrate a methodology for designing robust synthetic communities that use quorum sensing to control amensal bacteriocin interactions in a chemostat environment. We explore model spaces for two and three strain systems, using Bayesian methods to perform model selection, and identify the most robust candidates for producing stable steady state communities. Our findings highlight important interaction motifs that provide stability, and identify requirements for selecting genetic parts and tuning the community composition.