AI for Food Allergies project: open datasets and AI models (AlphaFold, ProtBERT, AllergenBERT) for allergen prediction

AI Impact Summary

The AI for Food Allergies initiative is building an open, community-driven research effort that integrates sequence-to-structure AI pipelines (AlphaFold, Boltz-1, ProtBERT, ESM-2, AllergenBERT) with open allergen datasets (SDAP, AllergenOnline, COMPARE, AlgPred 2) and data resources (PDBBind, DAVIS) to predict protein allergenicity and guide diagnostics. This framework enables in silico screening, rapid hypothesis generation, and multimodal analysis for diagnostics and therapeutics, potentially accelerating research timelines and enabling regulators and clinicians to access predictive insights sooner. Success hinges on data interoperability, licensing, model governance, and alignment with regulatory expectations, as well as maintaining data quality across the integrated sources and maintaining open access commitments.