Abstract Text The Mapping Component at Indiana University (MC-IU) will develop and implement at scale a socio-technical infrastructure that combines human expertise and machine learning algorithms to construct, visualize, and use the Human Reference Atlas (HRA). It will continue to lead the construction of ASCT+B tables (expert-curated, connected lists of anatomical structures, cell types, plus biomarkers based on standard ontologies) and associated reference object libraries that together define the evolving Human Reference Atlas. We will collaborate closely with the other mapping component to incorporate cell-by-gene references (e.g., Azimuth) and other cell-by-biomarker references and 3D reference objects as they become available. We will work with the HIVE and other teams to link the atlas to experimental data and scholarly paper evidence. We will work closely with 15 other international consortia to ensure the HRA meets user needs and supports driving use cases that advance biomedical research and clinical practice. In close collaboration with the HIVE and tissue data generating teams in HuBMAP and beyond, we will define and incrementally extend data formats and ontologies for querying, integrating, and sharing HRA data. We will incorporate modern machine learning and human-in-the-loop approaches to accelerate the pace at which raw tissue imaging data is converted into annotated and segmented tissue maps. Advanced spatial management will be used to keep track of millions of cells derived in tissue data, to support spatial registration of new tissue, and to enable efficient spatial queries and exploration in the HRA user interfaces. We will develop a vasculature-based common coordinate system to better capture human diversity and develop computational models of functional tissue units to link anatomical structure to function.

Public Health Relevance Statement The Human BioMolecular Atlas Program (HuBMAP) aims to create the very first computable, open Human Reference Atlas (HRA). This HRA will redefine our understanding of the human body by recovering multi-scale tissue organization—anatomical, histological, and molecular—at unprecedented resolution, through computational integration of diverse experimental measurements made by expert teams around the globe.