The European Food Safety Authority (EFSA) has published an external scientific report proposing an update, upgrade, and maintenance of the preDQ software tool, which predicts peptide binding to HLA-DQ2 and HLA-DQ8 proteins associated with susceptibility to celiac disease. The report, authored by Irini Doytchinova, Ivan Dimitrov, and Mariyana Atanasova and published on 15 July 2026, focuses on re-evaluating the docking-based models used in the tool. The preDQ tool is designed to evaluate the risk of novel proteins from genetically modified plants to trigger celiac disease, in line with EFSA's guidance on assessment of genetically modified plants.
The report outlines a project to update the datasets used to derive and validate the models, re-evaluate existing models with new data, derive new models, maintain the tool, provide feedback to users and stakeholders, publish models in peer-reviewed journals, and collaborate with EFSA experts. The current report specifically addresses the re-evaluation of docking-based models for peptide binding prediction to HLA-DQ2.5 and HLA-DQ8.1.
preDQ relies on five robust models, each producing five risk assessment outputs, developed using datasets of known binding and non-binding peptides. The models were derived by ligand-based and structure-based computational methods and validated by internal cross-validation and external datasets, with only the highest-performing models implemented in the tool.
The update aims to enhance the reliability of preDQ as a comprehensive, user-friendly tool for assessing the potential of novel proteins to cause celiac disease. This directly impacts EFSA's risk assessment processes for genetically modified plants, as well as applicants seeking market authorization for such products, who may face more accurate risk evaluations. Consumers could benefit from improved safety assessments of novel food proteins, while the scientific community gains access to updated models published in peer-reviewed journals. The project also ensures ongoing maintenance and user support, addressing potential concerns about the tool's long-term sustainability.