DIETxPOSOME: a FAIR database detailing food contaminants occurrence in selected foods

The DIETxPOSOME FAIR database provides detailed quantification of 73 contaminants, including 4 heavy metals, 18 polycyclic aromatic hydrocarbons (PAHs), 10 pesticides, 29 mycotoxins, and 12 heterocyclic aromatic amines (HAAs), in 16 food items across various food groups (cereals, vegetables, fruits, starchy roots, dairy products, nuts, meat, eggs, fish, legumes and vegetable oils). Data was obtained through a combination of literature extraction protocols and machine learning.

A list of pertinent research articles on food contaminants in globally significant food items was sourced from PubMed. Machine learning enabled the automatic filtering of these papers (https://zenodo.org/records/7826130), which were subjected to a manual evaluation. Subsequently, specific foods (wheat, maize, beef, cheese, pasta, potatoes, carrots, rice, bread, chicken eggs, peanuts, beans, cabbage, apple, olive oil, and salmon) were selected based on their comprehensive contaminant profiles. Information was extracted from 145 relevant articles of the 151 research articles pertaining to the selected food items. Following extraction, the data was confirmed and evaluated by three reviewers, adhering to stringent criteria. 

The food items in the DIETxPOSOME FAIR database can be integrated into complex dietary patterns in an experimental context, for example, aligned with the EAT-Lancet Commission Reference Diet for Healthy and Sustainable Food Systems. This framework allows the simulation of real-world dietary exposure or worst-case scenarios, facilitating comprehensive risk assessment of unavoidable food contaminants.