Multi-species microbial communities are widespread in natural ecosystems. When employed for biomanufacturing, engineered synthetic communities have shown increased productivity in comparison with monocultures and allow for the reduction of metabolic load by compartmentalising bioprocesses between multiple sub-populations. Despite these benefits, co-cultures are rarely used in practice because control over the constituent species of an assembled community has proven challenging. Here we demonstrate, in silico, the efficacy of an approach from artificial intelligence—reinforcement learning—for the control of co-cultures within continuous bioreactors. We confirm that feedback via a trained reinforcement learning agent can be used to maintain populations at target levels, and that model-free performance with bang-bang control can outperform a traditional proportional integral controller with continuous control, when faced with infrequent sampling. Further, we demonstrate that a satisfactory control policy can be learned in one twenty-four hour experiment by running five bioreactors in parallel. Finally, we show that reinforcement learning can directly optimise the output of a co-culture bioprocess. Overall, reinforcement learning is a promising technique for the control of microbial communities. In recent years, synthetic biology and industrial bioprocessing have been implementing increasingly complex systems composed of multiple, interacting microbial strains. This has many advantages over single culture systems, including enhanced modularization and the reduction of the metabolic burden imposed on strains. Despite these advantages, the control of multi-species communities (co-cultures) within bioreactors remains extremely challenging and this is the key reason why most industrial processing still uses single cultures. In this work, we apply recently developed methods from artificial intelligence, namely reinforcement learning combined with neural networks, which underlie many of the most recent successes of deep learning, to the control of multiple interacting species in a bioreactor. This approach is model-free—the details of the interacting populations do not need to be known—and is therefore widely applicable. We anticipate that artificial intelligence has a fundamental role to play in optimizing and controlling processes in synthetic biology.
The field of optimal experimental design uses mathematical techniques to determine experiments that are maximally informative from a given experimental setup. Here we apply a technique from artificial intelligence—reinforcement learning—to the optimal experimental design task of maximizing confidence in estimates of model parameter values. We show that a reinforcement learning approach performs favourably in comparison with a one-step ahead optimisation algorithm and a model predictive controller for the inference of bacterial growth parameters in a simulated chemostat. Further, we demonstrate the ability of reinforcement learning to train over a distribution of parameters, indicating that this approach is robust to parametric uncertainty.
Wastewater treatment plants are one of the major pathways for microplastics to enter the environment. In general, microplastics are contaminants of global concern that pose risks to ecosystems and human health. Here, we present a proof‐of‐concept for reduction of microplastic pollution emitted from wastewater treatment plants: delivery of recombinant DNA to bacteria in wastewater to enable degradation of polyethylene terephthalate (PET). Using a broad‐host‐range conjugative plasmid, we enabled various bacterial species from a municipal wastewater sample to express FAST‐PETase, which was released into the extracellular environment. We found that FAST‐PETase purified from some transconjugant isolates could degrade about 40% of a 0.25 mm thick commercial PET film within 4 days at 50°C. We then demonstrated partial degradation of a post‐consumer PET product over 5–7 days by exposure to conditioned media from isolates. These results have broad implications for addressing the global plastic pollution problem by enabling environmental bacteria to degrade PET.
1 October 2021 in Canadian Journal of Microbiology
The last two decades have seen vigorous activity in synthetic biology research and the ever-increasing applications of these technologies. However, pedagogical research pertaining to teaching synthetic biology is scarce, especially when compared to other science and engineering disciplines. Within Canada, there are only three universities that offer synthetic biology programs, two of which are at the undergraduate level. Rather than taking place in formal academic settings, many Canadian undergraduate students are introduced to synthetic biology through participation in the annual International Genetically Engineered Machine (iGEM) competition. Although the iGEM competition has had a transformative impact on synthetic biology training in other nations, its impact in Canada has been relatively modest. Consequently, the iGEM competition remains a major setting for synthetic biology education in Canada. To promote further development of synthetic biology education, we surveyed undergraduate students from the Canadian iGEM design teams of 2019. We extracted insights from these data using qualitative analysis to provide recommendations for best teaching practices in synthetic biology undergraduate education, which we describe through our proposed Framework for Transdisciplinary Synthetic Biology Education (FTSBE).
Quantitative characterizations of horizontal gene transfer are needed to accurately describe gene transfer processes in natural and engineered systems. A number of approaches to the quantitative description of plasmid conjugation have appeared in the literature. In this study, we seek to extend that work, motivated by the question of whether a mathematical model can accurately predict growth and conjugation dynamics in a batch process. We used flow cytometry to make time-point observations of a filter-associated mating between two {\em E. coli} strains, and fit ordinary differential equation models to the data. A model comparison analysis identified the model formulation that is best supported by the data. Identifiability analysis revealed that the parameters were estimated with acceptable accuracy. The predictive power of the model was assessed by comparison with test data that demanded extrapolation from the training experiments. This study represents the first attempt to assess the quality of model predictions for plasmid conjugation. Our successful application of this approach lays a foundation for predictive modelling that can be used both in the study of natural plasmid transmission and in model-based design of engineering approaches that employ conjugation, such as plasmid-mediated bioaugmentation.