consecutive_spatial_data

consecutive_spatial_data is a CLI tool for aligning histology images with spatial transcriptomics data. It is configurable via a config.toml file and designed primarily with Xenium spatial transcriptomics in mind, though it is compatible with any method supported by the SpatialData ecosystem.

Key Use Cases

• Align H&E-stained images to spatial transcriptomics data
  Automate the alignment of histology images acquired following spatial transcriptomics protocols.

• Align adjacent tissue sections for 3D reconstruction
  Register histological sections from adjacent tissue slices to reconstruct a 3D representation of the tissue alongside spatial transcriptomics data.

Features

• Automated Region of Interest (ROI) Detection

  • Automatically detect and segment multiple tissue sections per image.
  • Fine-grained control via the config.toml: skip poor-quality sections or align sections individually.

• Annotation Polygon Support

  • Align polygonal annotations (e.g., from Omero) to spatial transcriptomics data.
  • Align segmentations from other experiments (eg. align visium bins to xenium segmentations)

• Flexible Registration Options

  • Performs both rigid and non-rigid image registration.
  • Outputs are saved directly to a SpatialData object.

• Modular Backend Design

  • While Valis is the default backend, alternative registration methods can be integrated by implementing a process() method in a compatible class.

Getting Started

We recommend running consecutive_spatial_data inside the Valis Docker container for consistent results and dependency management. The container used during development was: cdgatenbee/valis-wsi:1.1.0

Example Usage

git clone github.com/callum-jpg/consecutive_spatial_data
cd consecutive_spatial_data
pip install -e .

consecutive_spatial_data register config.toml

An example configuration file can be found in config/example_config.toml.