OME-NGFF: scalable format strategies for interoperable bioimaging data

Biological imaging is one of the most innovative fields in the modern biological sciences. New imaging modalities, probes, and analysis tools appear every few months and often prove decisive for enabling new directions in scientific discovery. One feature of this dynamic field is the need to capture new types of data and data structures. While there is a strong drive to make scientific data Findable, Accessible, Interoperable and Reproducible (FAIR, 1), the rapid rate of innovation in imaging impedes the unification and adoption of standardized data formats. Despite this, the opportunities for sharing and integrating bioimaging data and, in particular, linking these data to other "omics" datasets have never been greater; therefore, to every extent possible, increasing "FAIRness" of bioimaging data is critical for maximizing scientific value, as well as for promoting openness and integrity. In the absence of a common, FAIR format, two approaches have emerged to provide access to bioimaging data: translation and conversion. On-the-fly translation produces a transient representation of bioimage metadata and binary data but must be repeated on each use. In contrast, conversion produces a permanent copy of the data, ideally in an open format that makes the data more accessible and improves performance and parallelization in reads and writes. Both approaches have been implemented successfully in the bioimaging community but both have limitations. At cloud-scale, those shortcomings limit scientific analysis and the sharing of results. We introduce here next-generation file formats (NGFF) as a solution to these challenges.