Spatial transcriptomics technologies — including Xenium, Visium, CosMx, and MERFISH — can now measure gene expression at near-single-cell resolution while preserving the spatial context of cells within tissue. This opens powerful new questions about how cell-cell interactions, tissue architecture, and microenvironments shape biological function, but analyzing these rich, heterogeneous datasets requires dedicated computational infrastructure that is accessible to domain experts without extensive computational experience.
The Bonham lab is developing SpatialOmics.jl, a Julia package that provides standardized, high-performance data structures for spatial transcriptomics analysis. Built around the SpatialData specification, it represents spatial points (e.g., individual transcripts), cell and nuclear segmentation shapes, and multi-scale images in a unified framework with full support for coordinate system transformations. SpatialOmics.jl takes full advantage of the julia package ecosystem to maximize performace and correctness. Lazy loading via DiskArrays.jl allows efficient analysis of datasets that are too large to hold in memory.
Core analysis operations are provided, including gene expression quantification per cell or region, point-in-polygon membership assignment, spatial density estimation, and k-nearest-neighbor networks. However, a crucial aspect of the design is that users should be empowered to enter or exit at any stage of the pipeline. SpatialOmics reads and writes in the SpatialData format or common tabular formats, allowing users to mix-and-match whatever tools are most appropriate for their science.