Machine learning and metabolic modelling of biofilms in the human gut

The commensal bacteria of the human intestine have been found to significantly affect
human health. They produce Vitamins B and K, important amino acids and essential
enzymes for carbohydrate fermentation that help improving digestion. In addition, they have
been linked to a variety of serious human diseases such as obesity, diabetes, autoimmune
disorders, allergies and even mental health. This has led to the development of a wide

range of probiotic products that include certain live microorganisms, such as Lactobacillus,
Bifidobacterium and Saccharomyces. In addition, gut bacteria often form biofilms, robust
structures that harbour mixed bacterial communities, attached to the intestinal epithelium.
The ability of such species to form biofilms is directly linked to their ability to colonise the
gut, which is essential for ensuring good efficiency of a probiotic product. This project will
provide a methodological advance in the field of biofilm metabolic modelling by
implementing multi-species modelling within a machine learning method. In the proposed
case study, it will also provide insights into how machine learning can be applied to find the
most optimal conditions for specific microbial communities in the human gut. The project
combines computer science, mathematics and biotechnology, and is intended to lead to a
long-term relationship among the three partners.