STP: single-cell partition for subcellular spatially resolved transcriptomics.

Installation

conda create -n STP python=3.7
conda activate STP
pip install -r requirements.txt

Usage

The required files include the h5ad file of spatial transcriptomics data and its corresponding pre-aligned nuclei-stained image.

The partition of single cell through spatial transcriptomics data is mainly done by the function called STP which includes following parameters:

• adata: h5ad file spatial transcriptomics
• img: Pre-aligned nulcei-stained image with spatial transcriptomics
• save_path: The path to save the results
• T: Initial temperature (the number of selected bins to extend the nuclei)
• T_min: Minimum temperature for ending the algorithm
• reduction_rate: Rate of reduction for T ( 0 < k < 1)
• neighbor_num: Number of neighbors for KNN
• alpha: Weight for distance energy
• beta: Weight for transcriptional similarity energy
• gamma :Weight for size energy
• thres: The threshold of nuclei segmentation

The output files include:

• Nuclei-segmentation image
• Single-cell partition image
• Connected component of partitioned cells
• The expression profiles of partitioned cells (index of cells is related to the label of the connected component of cell)

Example:

We supplied a demo for single-cell partition for the drosophila embryo. Finally, it would create a folder called example_result to save the output results introduced above.