RGAST

RGAST: A Relational Graph Attention Network for Multi-Scale Cell-Cell Communication Inference from Spatial Transcriptomics [paper]

This document will help you easily go through the RGAST model.

Dependencies

The required Python packages and versions tested in our study are:

pytorch==2.8.0
scanpy==1.11.5
scikit-learn==1.7.2
pyg==2.7.0
scipy==1.17.0
numpy==2.3.0
pandas==2.3.3

Installation

To install our package, run

git clone https://github.com/GYQ-form/RGAST.git
cd RGAST
pip install .

Usage

RGAST is a deep learning framework designed to infer multi-scale cell-cell communication (CCC) networks de novo from spatial transcriptomics (ST) data. RGAST integrates spatial proximity and transcriptional profiles using a relational graph attention mechanism. This approach allows RGAST to dynamically learn context-specific signaling patterns and reconstruct CCC networks without prior knowledge of ligand-receptor pairs, effectively capturing both local and global communication patterns. Besides, RGAST is also a versatile tool for many downstream ST analysis:

• spatial domain identification
• spatially variable gene (SVG) detection
• cell trajectory inference
• reveal intricate 3D spatial patterns across multiple sections of ST data

Tutorial

We have prepared several basic tutorials in https://github.com/GYQ-form/RGAST/tree/main/tutorial. You can quickly hands on RGAST by going through these tutorials.

Analysis

To enhance the reproducibility of this study, we deposited all the custom code at Zenodo repository for running RGAST used in the paper. A comprehensive README file has also been provided for easy using of these custom scripts.