Using digital twins to predict the evolution of food microbiota during vegetal fermentation

Context and challenges

Drinks based on fermented vegetable juices are becoming increasingly popular for their taste, nutritional benefits and potential probiotic qualities, given the multitude of microbial species involved in their production.

The control of continuous fermentation during production is a major issue for the industry because it is essential to the achievement of consistency in organoleptic and sanitary quality while controlling costs. Optimisation of the industrial production of fermented juice calls for both improved understanding of the underlying mechanisms of fermentation and control over the microbiota involved in the process.

The FermenTwin project plans to develop a digital twin that will model, influence and predict the behaviour of the microbial community during the fermentation of carrot juice , through real-time sequencing of its microbiota.

This project, which operates at the interface between microbiology, robotics and mathematical modelling, will enable in silico monitoring of the microbial community and its metabolism in order to predict its evolution in response to biotic and abiotic shifts.

 Goals

The FermenTwin project seeks to ensure that ‘proper fermentation’ is achieved in the model under study, a goal that requires us to negotiate a number of challenges concerning, on the one hand, our capacity to successfully design the biological model, influence the experimental model, and create a mathematical model able to describe how the phenomenon functions as a system and, on the other, our ability to develop effective decision-making processes to influence fermentation.

 To achieve its objective, the project has been set up in 4 stages:

• Preliminary work : deployment of monitored mini-bioreactors based on prototypes, design of the reference microbial community for the production of fermented carrot juice.
• Measurement of bacterial community dynamics using Oxford Nanopore sequencing (real-time sequencing).
• Modelling of ecosystem dynamics to enable decision making: defining a reference dynamic that describes how the system works and modelling the impact of biotic and abiotic disturbances on this reference. A decision model will then be developed to enable the digital twin to influence the experimental model during fermentation, thereby restoring the reference dynamic.
• Assessment of automation of all stages, in particular by identifying critical points in software-biology-machine interactions.

Ultimately, the use of a digital twin to control the continuous fermentation of a liquid plant matrix will provide new opportunities for food technologies to stabilise and optimise processing.

Contact-coordination

Project participants

INRAE structures

Non-INRAE partners