Introduction

Learn about TileDB-SOMA’s support for spatial omics datasets.

Spatial omics technologies, which enrich traditional genomic assays with spatial context, are rapidly advancing and becoming more accessible to researchers. These technologies present researchers with new opportunities to study the spatial organization of tissues and cells, enabling a deeper understanding of complex biological systems. However, the analysis of spatial omics data also presents new challenges around data management and accessibility. TileDB-SOMA has addressed these challenges by extending the core SOMA data model to add native support for spatial data elements.

This extension, designed in collaboration with the Chan Zuckerberg Initiative (CZI), adds new capabilities for representing point clouds, geometries (that is, shapes), and multiscale images within the SOMA framework. It also introduces the SOMAScene, a specialized collection type for organizing spatially resolved data elements within a single coordinate space, and offers mechanisms for transforming coordinates between different representations. This spatial support augments the existing SOMA model, while retaining its core strengths for handling annotated matrices.

Release status

TileDB-SOMA’s spatial data support is a beta feature, introduced in version 1.15.0. Features and documentation are subject to change. Contact us if you have any feedback or suggestions you would like to share.

The initial beta release focuses on the following:

• Supported Assays: Spot-based assays, including 10x Genomics Visium and Slide-seq.
• Core Data Types: Foundational classes for storing multiscale images (MultiscaleImage), point clouds (PointCloudDataFrame), and geometries (GeometryDataFrame). These data types are associated with a CoordinateSpace.
• Basic Querying: Reading and slicing data within these new types.
• Coordinate Transformations: Support for transformations including IdentityTransform, ScaleTransform, and AffineTransform to enable converting coordinate systems between a SOMAScene and spatial objects stored inside it.
• Spatial Queries: Bounding box queries on multi-resolution images, point cloud, and geometry data.
• Spatial Data Interoperability: Support for exporting to SpatialData objects.

Current Limitations and Future Plans:

• Support for image-based assays (such as 10X Xenium or Nanostring CosMx) is planned for future releases.
• Higher-level APIs for experiment-level querying and more advanced coordinate transform.
• Full R support with support for Seurat and Bioconductor.

Section organization

The rest of the Single-Cell Spatial section is as follows:

• Foundation: This has all the background information and internal mechanics of the TileDB-SOMA spatial omics support.
• Tutorials: This is a series of tutorials covering the existing spatial functionality.

Refer to the API Reference doc for fast lookups on API usage for spatial functions in TileDB-SOMA.