NMRhub: An NMR Data Ecosystem Spanning the Complete Data Lifecycle

The emergence of reporting standards represents a tremendous stride toward reproducible science, but significant challenges remain. Preserving data provenance is difficult, from generation through processing and analysis to publication. The accuracy of annotations diminish the longer the annotation is separated from the defining event. Manual annotations are often inaccurate or inconsistent. Federation of data across data resources remains challenging. NMRhub was conceived to address these challenges, and to facilitate the application of machine learning (ML) to unlock latent knowledge. NMRhub links the NSF Network for Advanced NMR (NAN), the NMRbox platform, and the Biological Magnetic Resonance Data Bank (BMRB) to create an integrated ecosystem that sustains data from the instrument source through public deposition, and facilitates surveys, meta-analyses, and ML. NAN harvests data automatically from connected spectrometers and stores it in a secure archive, parses instrument-provided metadata and transmits data to the user NMRbox account. NMRbox virtualizes and archives the complete software environment used for processing and analysis, provides high performance computation resources including access to Open Science Grid, and tools for assembling BMRB depositions. BMRB, a Core Archive of the World-wide Protein Data Bank (wwPDB), curates, annotates, and provides persistent identifiers for all types of biomolecular NMR data. NMRhub provides single sign on across the constituent platforms, ML-enhanced search of assets across the three platforms, as well as across curated external resources (e.g. PubMed, domain-specific YouTube channels, funding databases), and tools for data transfer among the linked resources. The NMRhub ecosystem spans the complete NMR data lifecycle.