domains · spatial omics
tissue atlases as tensors
Spatial omics produces exactly the kind of data the array model was built for: high-dimensional measurements on a physical grid. Earthmover stores it at atlas scale without flattening it into tables.
the data
What we store
Spatial transcriptomics
Expression measured in place, keeping the spatial coordinates alongside the counts.
Multiplexed imaging
Many-channel tissue imaging where every marker is a band of one array.
Cell-by-gene matrices
Large sparse and dense expression matrices with their full annotation.
Whole-slide imagery
Multi-resolution histology pyramids stored beside the molecular data they register to.
on the platform
What Earthmover adds
-
Channels as a dimension — add markers or genes without restructuring the dataset.
-
Atlas scale — many samples in one logical dataset, streamed from object storage.
-
Versioned pipelines — segmentation and annotation runs are commits you can compare and revert.