Applying NGS in the epidemiology of foodborne infections

Each year, nearly one in ten people globally are affected by foodborne diseases (FBDs) with the greatest burden falling on sub-Saharan Africa. Children under the age of five are disproportionally affected. In many low- and middle-income African countries (LMICs), the limited capacity for surveillance and monitoring makes it difficult to understand the true scope of FBDs, identify their causes, and respond effectively to outbreaks. This gap in epidemiological knowledge impedes progress in food safety and surveillance systems, and places a strain on healthcare systems in these countries.

This project was initiated in response to these challenges, with the aim of exploring how genomic technologies such as whole genome sequencing (WGS) and metagenomic sequencing can be applied to improve the detection, attribution, and understanding of FBDs in African LMICs. By evaluating the performance and feasibility of these methodologies in African LMICs, the objectives were to identify sources and risk factors for FBDs, assess the use of metagenomic sampling for surveillance, develop attribution methods using metagenomic data, and explore barriers and strategies for improving FBD monitoring.
These topics are addressed through four manuscripts.

Manuscript I evaluated the use of WGS- and metagenomic sampling from sewage, wastewater, and other sources to support FBD surveillance in African LMICs. The study found high genetic diversity among E. coli isolates and highlighted the value of combining WGS and metagenomics. Metagenome-assembled genomes showed strong consistency with WGS data, supporting metagenomics as a feasible complement to isolate-based methods in settings where culturing and isolation are challenging to perform due to limited resources.

Manuscript II demonstrated that targeted genomic surveillance using WGS can detect clusters and identify likely sources of Salmonella infection, even with limited data. Phylogenetic analysis and source attribution of 68 isolates from humans, animals, and the environment in Ethiopia revealed three clusters, with most human cases linked to bovine sources. Despite sampling constraints, the study showed that WGS-based source attribution can provide meaningful evidence to inform public health responses in LMIC settings. At the same time, it highlighted the need for broader, more representative surveillance to improve accuracy in these settings.

Manuscript III introduced a novel approach for source attribution based on low-abundance E. coli strains detected in metagenomic samples from children with diarrhea. By combining the StrainGE toolbox with a Bayesian modeling framework, the study demonstrated how strain-level matches could be used to link infections to livestock reservoirs, particularly bovine and poultry. The method showed a promising possibility for identifying transmission routes with non-conventional methods such as metagenome sequencing.

Manuscript IV applied a systems thinking framework to identify key challenges and opportunities for improving FBD surveillance in African LMICs. The study mapped out structural and behavioral components of existing systems and highlighted critical leverage points that could strengthen surveillance when effectively addressed. Elements such as public trust, food safety compliance, and cross-sectoral data sharing were shown to influence essential system functions like case reporting, collaboration, and enforcement. To create more resilient surveillance systems, the findings point to the need for long-term strategies focused on trust-building, institutionalizing compliance, and promoting transparency, alongside short-term, practical interventions that are locally feasible.

In summary, the research conducted in this PhD highlights the potential of integrating genomic tools into FBD surveillance in African LMICs. However, technological advances alone are not sufficient. Sustainable improvements will require long-term investment in local laboratory infrastructure, workforce training, and robust data systems. Ensuring that future surveillance is grounded in local expertise rather than short-term external support is also crucial. Moving forward, research should prioritize comparative studies of metagenomic and traditional methods to assess their cost-effectiveness and practicality in LMIC settings. Importantly, LMICs should avoid replicating surveillance models from high-income countries and instead develop systems that suit their countries while taking advantage of new technologies.