Tripso (Transformers for learning Representations of Interpretable gene Programs in Single-cell transcriptomics) is a self-supervised framework that decomposes cellular states into multiple gene program (GP) embeddings. Tripso learns contextualised gene embeddings within each GP, quantifies gene-level contributions to each GP, and the influence of each GP on global cell identity.

Resources

• Installation, tutorials, and API reference: tripso documentation (built with Sphinx; theme provided by Read the Docs).
• Bug reports: use GitHub Issues.
• Contributing: pull requests are welcome. The repo uses checks for PEP 8 and static typing; install pre-commit and run pre-commit install before committing. For building the docs locally, see Building the documentation.

Citation

If you use Tripso in your research, please cite our work (citation to be added).