PhD defence by Alexander Gmeiner

Supervisors:

• Principal supervisor: Senior Researcher Pimlapas Leekitcharoenphon
• Co-supervisor: Senior Researcher Patrick Murigu Kamau Njage
• Co-supervisor: Professor Frank Møller Aarestrup

Examiners:

• Associate Professor Håkan Vigre, DTU Food
• Associate Professor Lukasz Krych, University of Copenhagen
• Assistant Professor Sophia Johler, Institute for Food Safety Hygiene, University of Zurich

Chairperson at defence:

• Senior Researcher Rolf Sommer Kaas

Resume

Foodborne disease pathogens continue to threaten at-risk populations all around the world. Over 350,000 foodborne disease cases are reported yearly in the European Union alone. Among the most common causes of foodborne disease are bacterial pathogens, such as Listeria monocytogenes.

L. monocytogenes is a concerning pathogen that is mainly transmitted through food and can cause severe diseases in susceptible communities, such as meningitis and sudden pregnancy terminations. However, not all L. monocytogenes isolates are the same. Like humans, L. monocytogenes isolates can have different characteristics and abilities even though they belong to the same species. Accordingly, to better understand the potential threat of a given L. monocytogenes isolate, it is essential to characterize their abilities in as much detail as possible. The detailed description of the potential risks is critical when food or food production environments are contaminated with L. monocytogenes. In such cases, risk management professionals might ask important questions like: “How dangerous are the
found pathogens?” and “How do you best eliminate them?”

This PhD research aimed to help unravel these questions by developing predictive machine learning models that use sequencing data to predict the virulence potential and the tolerance to disinfectant of L. monocytogenes isolates. All in all, we were able to train highly accurate ML models that can predict the virulence and tolerance of L. monocytogenes in great detail. Apart from that, we were able to identify genomic features (i.e., genes) that are potentially linked to virulence or tolerance mechanisms. To facilitate the translation of this academic research into practice, we developed a user-friendly web tool called ListPred that can be used publicly and free of charge to analyse L. monocytogenes sequencing data.

In the future, predictions from ListPred could potentially be used by risk managers in the food industry to better characterize L. monocytogenes isolates, which could help them make decisions about risk mitigation procedures. The disinfectant tolerance predictions could also guide sanitation plans to eliminate and prevent L. monocytogenes contamination as efficiently as possible. In general terms, implementing sequencing technologies and machine learning algorithms has immense potential to improve food safety. The development of easy-to-use tools that help to analyse and interpret sequencing and machine learning data can greatly promote the use of these beneficial technologies in the food industry.