Nanosafety data made interoperable using semantic modeling and linked-data knowledge graphs

Achieving data interoperability is a critical challenge in the increasingly complex landscape of the nanosafety field, where ensuring the safe use of nanomaterials is of great importance. One significant challenge lies in the diversity of experimental approaches, measurement techniques and exchange formats employed in nanosafety research. Fortunately, semantic modeling coupled with linked-data knowledge graphs emerges as a powerful solution. Semantic modeling involves structuring data in a way that adds meaning and context to the information, facilitating better harmonization and standardization. Linked-data knowledge graphs take this a step further by establishing relationships between diverse datasets and their metadata. The semantic model presented in this work adopts several ontologies to describe the datasets and their metadata. For example, DCAT and VoID were used to describe metadata, NPO for nanomaterial entities and BAO and eNanoMapper for bioassays and experimental conditions. The model captures two types of assays, toxicity assays and gene expression assays. This approach utilizes the RDF Mapping Language (RML) to represent the semantic model as reusable mapping rules. Then, the knowledge graph can be explored using SPARQL query language to answer queries such as finding gene expression patterns at concentrations where a nanomaterial is deemed toxic. This semantic approach is essential for advancing our understanding of nanomaterials' safety profiles. It allows for better understanding and seamless data integration and exchange across different applications. Moreover, it inherently complies with the FAIR principles (Findable, Accessible, Interoperable and Reusable), thus making the data more accessible and reusable for the community.