Improving annotation, access, and comparison of nutritional composition data for underutilized crops

Open and interoperable data infrastructures are essential to advancing food and nutritional security research, yet few data management systems are specifically designed for crop nutritional datasets that support seamless querying, visualization, and comparison. Existing data sources are often syntactically and semantically heterogeneous, creating substantial barriers to interoperability and data reuse, and ultimately limiting the translation of research into sustainable agricultural innovation. To address this gap, we developed an ontology-based data access and integration (OBDI) framework that harmonizes publicly available, heterogeneous plant nutritional datasets within a unified semantic structure. By annotating datasets using established plant science ontologies, we enhanced their Findability, Accessibility, Interoperability, and Reusability by providing a scalable mechanism for virtual data integration via a knowledge graph enriched with domain semantics. Our workflow enables consistent comparison and computational reasoning across crop nutritional composition data, supporting both machine and human interpretation. The integration of compositional datasets for underutilized crops (UCs) alongside major crops allows for the identification of genetic resources that provide enhanced nutritional outcomes, fostering evidence-based diversification strategies. We outline how this approach may facilitate data integration and assist in the wider adoption and utilization of UCs. More generally, this open, ontology-driven approach highlights how investment in standardized, FAIR-aligned data infrastructure has the potential to accelerate interdisciplinary collaboration across plant sciences, nutrition, and policy. We demonstrate the concepts and specific methodologies for establishing semantic relationship between diverse datasets useful in operationalizing heterogeneous crop trait phenotyping knowledge bases. Furthermore, we discuss how the integration of ontology-based data integration (OBDI) with machine learning enables intelligent, accessible trait data management, thereby enhancing the adoption and advancement of AI-driven decision-making.