The Common Fund (CF) research initiative has generated a wealth of data that can provide vital context, origin, and, in some instances, quantitative inferences for biomarkers. However, the systematic harmonization and organization of biomarker data, as well as their connections to CF data, remain in an early stage and is currently the focus of the year-long Common Fund Data Ecosystem (CFDE) Biomarker-Partnership project that aims to develop a working biomarker data model. The proposed BiomarkerKB project aims to refine and populate the biomarker data model through a close and dynamic external partnership with the NCI-supported Early Detection Research Network (EDRN) with built-in community input mechanisms. The initial focus is on refining our current biomarker data model using EDRN's cancer biomarker data and knowledge. Initially focusing on cancer will allow us to limit our scope while retaining the ability to evaluate a variety of data types and therefore ensure extensibility of the model as new data types and technologies emerge. The Minimal Viable Product (MVP) will include persistent biomarker identifiers, linked data, connections to recognized standards and ontologies, downloads/APIs, and data access through interfaces for biomarker explorations. This data model will serve as the cornerstone for AI-ready datasets, machine learning-based biomarker prediction models, and biomarker knowledge graphs. The scientific use case the project proposes to support is the ability to explore molecular biomarker-related knowledge for most prevalent cancers at a systems level, categorized by biological functions through mapping to key ontologies, pathways, biomolecular data (glycans, proteins, genes, metabolites) and Electronic Health Record (EHR) terms and tests. Example biomarkers (including non-molecular ones that are of interest to Data Coordinating Centers (DCCs)) for other diseases will also be considered to ensure the comprehensiveness and robustness of the data model. This project promises to enrich our understanding of the translational health record and intervention space, revolutionizing the way we approach diverse diseases, clinical assays, molecular mechanisms, and disease classifications. The potential benefits extend to our partners in the EDRN and the broader CFDE community, underscoring the real-world significance of biomarkers across the medical and research landscape.

The proposed project will establish a structured representation of biomarkers of clinical interest, with links to related knowledge models (e.g. genes, proteins, glycans, metabolites, disease, phenotype, cell, tissue, and EHR codes). Semantic modeling of biomarkers will empower novel comparisons across related medical domains, Common Fund datasets, and make biomarker information findable, accessible, interoperable, and reusable (FAIR).