Spots-In-Space

A Python based scalable pipeline to apply the Cellpose segmentation algorithm to large, subcellular resolution spatial transcriptomics datasets. This tool provides wrapper functions to tile large images, distribute the segmentation jobs on a SLURM cluster, and stitch the results back together. We also provide the option to create an mRNA density image from the spot table as a label for the cytosolic regions.

✨ Features

• Tiling & Stitching: Automatically splits large images into tiles for processing and reassembles the final segmentation masks.
• mRNA Density Image: Generates mRNA density image from spot table.
• SLURM Integration: Distributes segmentation tasks across cluster nodes for high-throughput processing.
• Comprehensive Outputs: Generates standard cell-by-gene tables, metadata files, and cell outlines in GeoJSON format.
• Built-in Visualization: Includes plotting functions to easily visualize cell boundaries overlaid on the original image data.

Installation

Create a conda environment in which to install spots-in-space:

conda create -n sis python=3.12
conda activate sis

We recommend cloning this repo and installing via pip:

git clone https://github.com/AllenInstitute/spots-in-space.git
cd spots-in-space
pip install ".[cellpose]"

Should you be interested in a distribution of spots-in-space without cellpose, it can be installed as such:

git clone https://github.com/AllenInstitute/spots-in-space.git
cd spots-in-space
pip install .

Outputs

This pipeline generates:

1. Cell-by-Gene Table: A matrix with cells as rows and genes as columns
2. Metadata: A file containing descriptive data for each segmented cell
3. Cell outlines: In the form of a geojson file

Level of support

We are planning on occasional updating this tool with no fixed schedule. Community involvement is encouraged through both issues and pull requests.