Listeria Genome Identification Using DNABERT Embedding With LightGBM and SHAP-Based Explainable Classification

The Quick Summary

Listeria is a tiny germ that can make people very sick, especially from food. Finding this specific germ, called Listeria monocytogenes, quickly and accurately in food is super important to keep everyone safe and stop sickness from spreading. Right now, the ways we find it are often slow, hard to do, or use computer programs that don’t explain how they made their decision. This study wants to make a new, better way to find Listeria germs using special computer tools. It uses advanced computer code to look at the Listeria’s DNA very closely. This new method will be faster and clearer, so we can understand exactly why the computer thinks Listeria is present. This helps food scientists keep our food safer.

Practical Implications

This study has significant practical implications for food science, particularly in enhancing food safety protocols. By offering a faster, more accurate, and crucially, explainable method for identifying Listeria monocytogenes at the whole-genome level, food producers and regulatory bodies can react to contamination events with greater speed and precision. The interpretability provided by the SHAP-based classification means that food safety experts can understand why a particular sample is flagged, allowing for targeted interventions and improved confidence in the diagnostic results. This can lead to a reduction in the incidence of foodborne Listeria outbreaks, minimize product recalls, and ultimately bolster consumer trust in the food supply chain. Furthermore, its efficiency could integrate seamlessly into routine surveillance programs, making comprehensive genomic analysis more accessible.

Potential Use in Indonesia

In Indonesia, with its vibrant traditional markets, street food culture, and tropical climate, rapid and explainable detection of Listeria monocytogenes would be invaluable. This technology could significantly enhance food safety in ready-to-eat foods and fresh produce often sold with limited cold chain management, reducing the risk of foodborne illnesses among the population. Implementing such a system could empower local authorities to quickly identify and mitigate contamination sources, safeguarding public health across diverse food supply chains.

Original Abstract

Prompt and accurate identification of Listeria monocytogenes at the whole-genome level is essential for food safety surveillance and outbreak prevention, yet existing culture-based, PCR, and next-generation sequencing (NGS) workflows are either slow, labor-intensive, or rely on opaque machine learning models with limited interpretability. This study proposes an explainable genomic classification framework that couples transformer-based DNA embeddings with gradient boosting to distinguish……

Read the full peer-reviewed publication here