Marija Gavrović-Jankulović*
University of Belgrade – Faculty of Chemistry
mgavrov [at] chem.bg.ac.rs
Abstract
Predictive modeling of immune epitopes has become an important component of computational allergology, enabling systematic analysis of allergenicity and cross-reactivity of proteins. This work presents an integrated in silico framework for the identification of B- and T-cell epitopes in model of food and inhalant allergenic proteins, combining sequence homology analysis, motif-based approaches, and machine learning–driven predictors. Linear B-cell epitopes are identified using sequence-derived features and propensity scores. In parallel, T-cell epitopes are prioritized based on predicted binding to major histocompatibility complex (MHC) class II molecules, incorporating peptide binding affinity and population coverage analyses.
In this study key focus is on the identification of conserved epitope patterns across homologous proteins, enabling the detection of cross-reactive allergens and panallergen families. Such approaches facilitate the interpretation of IgE cross-reactivity observed between phylogenetically related and unrelated allergen sources. The predictive approach further supports the rational design of hypoallergenic variants by selectively modifying IgE-binding regions while preserving T-cell immunogenicity. In addition, large-scale screening of protein datasets enables the identification of novel allergen candidates based on epitope signatures associated with sensitization risk.
Integration with experimental IgE-binding and immunological validation enhances the robustness and translational relevance of these predictive approaches.
Keywords: computational allergology, protein allergenicity prediction
Acknowledgement: MGJ acknowledge financial support to the Ministry of Science, Technological Development and Innovation (contract numbers: 451-03-136/2025-03/200168, 337-00-216/2023-05/133), the European Union’s Horizon Europe research and innovation programme under the Marie Skłodowska-Curie grant agreement No 101072377.