Uncovering directionally and temporally variable genes with STAVAG

Tutorials

• Case I: STAVAG on 2D cSCC data
• Case II: STAVAG on 3D planarian
• Case III: STAVAG on STARmap 3D cortex
• Case IV: STAVAG on mouse myocardial infarction progression data
• Case V: STAVAG on mouse embryonic development data

API References

• API References

---

Overview of STAVAG

STAVAG can handle spatial transcriptomics data for single, 3D, or multiple slices coupled with temporal information. STAVAG takes spatial transcriptomics data as input and then fits the spatial or temporal direction of spatial data with gene expression using a gradient boosting tree for regression. STAVAG calculates the gene contribution score along any given direction or temporal progression for each gene and identifies directionally variable genes (DVGs) and temporally variable genes (TVGs) for different scenarios.

Installation

Set up conda environment for STAVAG:

conda create -n STAVAG python==3.9

activate STAVAG from shell:

conda activate STAVAG

you can install the important Python packages used to run the model are as follows:

pip install "scanpy[leiden]"
pip install lightgbm
pip install numpy
pip install matplotlib
pip install scikit-learn
pip install scipy

now you can install the STAVAG Python package as follows:

pip install STAVAG

Citation

Shen Q, Gai K, Li S, et al. Uncovering directionally and temporally variable genes with STAVAG. bioRxiv, 2025. doi:10.1101/2025.09.02.673732