Dual schema of allergens reveals molecular and evolutionary signatures

Allergies affect billions of people worldwide, posing a substantial challenge to global health because of its elusive molecular determinants. Leveraging high-quality protein structures, we revisit why allergenic proteins trigger allergies, whereas some of their structural analogues do not. We categorize allergenic protein folds into similar (SAP) and dissimilar (DAP) folds on the basis of their similarity to human proteins, establishing a dual schema that reveals distinct molecular and evolutionary patterns. Here, we show that compared with structurally similar, nonallergenic protein (SNAP) folds, SAP folds are ubiquitous across kingdoms and rich in B- and T-cell epitopes. DAP folds map to a unique sequence–structure–function space and are evolutionarily underrepresented and taxonomically restricted. The persistent divergence of DAP folds from animal proteins suggests their intermittent immune encounters. In contrast, ubiquitous SAP folds require continuous immune sorting from their SNAP analogues. This dichotomy provides a new lens for studying the evolution of host immune responses in allergic reactions. Our classifier, AllerX, is trained on these patterns and predicts unseen APs with state-of-the-art accuracy. This work, which links the structural, immunological, and evolutionary aspects of allergenic proteins, establishes a novel framework for enhancing diagnostics and food safety. Significance Allergies, a global health challenge, cause life-threatening immune reactions. Our theory links protein foreignness to immunogenicity, but its structural definition based on similarity to human proteins remains undefined. Utilizing massive protein structure resources, we grouped allergenic protein folds into similar (SAP) or dissimilar (DAP) to human proteins. We reveal strong contrasts in molecular and evolutionary signatures: DAP folds are taxonomically restricted but have biased, source-specific distribution among kingdoms. SAP folds show prominent, orientation-specific B-cell epitopes compared to their non-allergenic structural analogs. We also show fold-independent T-cell epitope linkage between SAP and DAP folds as a converged allergenicity signal. This dichotomy enhances cross-reactivity tracing in food safety incidents and establishes a foundation for studying host-factor evolution of protein allergy.