Extended Data Fig. 2: Schematic representation of the SpatialData aggregation operations.

Aggregation operations are the foundation to flexibly transfer quantifications and annotations across modalities when conducting multimodal analyses. SpatialData enables the aggregation (also referred to as accumulation in image processing) of data stored in any SpatialElement into any set of target geometries or masks. Example applications include count aggregation of the number of single molecules for a specific gene within polygon geometries representing cells. Similarly, molecule counts within image masks representing the cytoplasm of the cells. Another example is averaging cell gene expression within a given anatomical region (see also main text Fig. 2). Predefined aggregation operators (count, sum, mean, standard deviation) can be applied to any SpatialElement. Additionally, SpatialData supports the definition of user-specified aggregation operators. Leveraging common coordinate systems, aggregation operations can be applied to collections of datasets, including across datasets with different spatial scales and/or partially overlapping datasets. Tutorials on how to use the aggregation system are available as part of the SpatialData online documentation (https://spatialdata.scverse.org/en/latest/tutorials/notebooks/notebooks/examples/aggregation.html).