Installation

  1. Create and activate a conda environment with python 3.12

conda env create spamv python==3.12
conda activate spamv

  1. (Optional) If you want to apply our algorithm to large datasets (with more than 10,000 spots), please make sure you have installed the pyg-lib package.

pip install pyg-lib -f https://data.pyg.org/whl/torch-${TORCH}+${CUDA}.html

where

  • ${TORCH} should be replaced by either 1.13.0, 2.0.0, 2.1.0, 2.2.0, 2.3.0, 2.4.0, 2.5.0, 2.6.0, or 2.7.0

  • ${CUDA} should be replaced by either cpu, cu102, cu117, cu118, cu121, cu124, cu126, or cu128

  1. Then you can install our package as follows:

pip install spamv

Tutorial Instruction

We provide two jupyter notebooks (Tutorial_simulation.ipynb and Tutorial_realworld.ipynb) to reproduce the results in our paper. Before you run them, please make sure that you have downloaded the simulated data and/or real-world data from our Zenodo repositoy.