Cryo-EM Data at Scale: How Collaborative Models Arose Around the World #RSAA25

Abstract: Cryo-Electron Microscopy (Cryo-EM) —the electron imaging of biomolecules flash-frozen in their near-native state— is now delivering direct atomic level views of structures that previously could not be resolved. This view provides immediate mechanistic insight, often clarifying biological processes that had long been debated. But with this mass of atomic level data comes a massive data challenge. Cryo-EM instruments can generate terabytes of imaging data per day, and this data volume cannot be handled in isolation. To separate the molecular wheat from the chaff, skilled instrument scientists are critical to achieving high resolution; biomolecular domain experts triage live data to identify what’s worth interpreting; and research computing provides the processing power to reconstruct these large datasets into 3D maps. The Free and Open Source Software (FOSS) community has demonstrated several models for open collaboration that thrive in environments where code sharing and reproduction is near-instantaneous. But as the scale of biodata grows, the demands for storage and compute will increasingly necessitate bringing resources and expertise to the data sources — not the other way around. In this talk, I’ll share experiences coordinating across facilities with different approaches and assumptions to meet shared Cryo-EM needs. I’ll then turn to highlight some international efforts to address the Cryo-EM data and modelling burden — including software packaging and distribution models, as well as inter-facility and even inter-university infrastructure sharing approaches, such as the multicountry Nordic CryoNET initiative.