T-GRAASP: Reconstructing Spatial Context in Highly Heterogeneous Tissues

T-GRAASP is a computational framework designed for reconstructing the spatial organization of cells and analyzing cellular interactions in highly heterogeneous tissues, based on spatial transcriptomics data. T-GRAASP leverages graph-based modeling to accurately infer spatial adjacency, identify spatial domains, and dissect gene interaction networks across various tissue types.

Make sure you have Miniconda or Anaconda installed.

git clone https://github.com/caofy0307/T-GRAASP.git
cd T-GRAASP

If an environment.yml file is provided (recommended for full reproducibility):

conda env create -f environment.yml
conda activate t-graasp

This project provides detailed tutorial notebooks to help you get started quickly.
After installing the environment, simply open the following Jupyter notebooks in the repository:

• Pretrain.ipynb
  This notebook demonstrates the pretraining workflow of the T-GRAASP model. Use it to learn how to prepare your data and pretrain the model for spatial transcriptomics analysis.

• Finetuned.ipynb
  This notebook shows how to fine-tune the pretrained T-GRAASP model on your own datasets and analyze results, including visualization and downstream analyses.

• Downstream.ipynb
  This notebook provides downstream biological analysis and rich visualization examples, helping you to interpret and explore the results from T-GRAASP in a biological context.

1. Activate the Conda environment:

   conda activate t-graasp

2. Launch Jupyter Notebook:

   jupyter notebook

3. Open Pretrain.ipynb, Finetuned.ipynb, and Downstream.ipynb in your browser and follow the step-by-step instructions.
   These notebooks are located in the root or notebooks/ directory of this repository.

The notebooks include comments and code cells that are ready to run.
You can adapt them for your own data or use the provided example datasets.

For questions, bug reports, or collaborations, please open an issue.