Ouyang, W., Bai, J., Singh, M.K. et al.
Single-molecule localization microscopy (SMLM) has matured into one of the most widely used super-resolution imaging methods and has been used to address a broad spectrum of biological research questions1. This success has inspired the community to develop numerous computational techniques to extract localizations from raw images or turn them into biologically meaningful quantities1,2,3. The development of further analytical methods could greatly benefit from easy access to SMLM data generated worldwide. This is especially true for machine learning approaches and notably deep learning, whose performance hinges strongly on the amount of training data. However, despite the vast number of SMLM studies1, the overwhelming majority of SMLM data remains inaccessible to the community. Repositories such as Figshare or Zenodo, or the added-value repository IDR4,5, are generic in purpose and not optimally suited for gathering, visualizing and exploiting SMLM data in a manner consistent with the FAIR principles (findability, accessibility, interoperability and reusability)6. An important impediment for sharing SMLM data is that each super-resolution image is built by computing molecular coordinates from many thousands of raw, low-resolution images (Fig. 1a), which often total many gigabytes in size. Even localization files, albeit much smaller than the raw data, are usually too large to be sent by e-mail.