Cafe: An integrated platform for exploring cell fate

Cafe (Cellular Fate Explorer) is a modular framework to study cellular dynamics based on single-cell RNA-seq data. It provides an integrated platform for inferring, visualizing and benchmarking cell fate trajectories.

Framework

Key Applications

• Trajectory Inference: Infer cell fate trajectories using various backend methods (including Python-based, Docker-based, and Conda-based backends).
• Visualization: Visualize trajectories, embeddings, pseudotime, and velocity fields with high-quality plots.
• Benchmarking: Compare different trajectory inference methods using comprehensive metrics including topology, cluster, and feature importance.
• Data Management: Manage complex single-cell data structures with FateAnnData, extending the capabilities of AnnData for trajectory analysis.

Documentation

Comprehensive documentation is available at here, including installation, tutorial, API and so on.

Interactive web launch

Cafe can open a saved .h5ad file in native CELLxGENE or in the CAFE-enabled cafe-cellxgene single-dataset viewer.

Launch native CELLxGENE:

cafe launch-cellxgene dataset.h5ad --port 5005

Launch with the CAFE plugin:

cafe launch-cellxgene dataset.h5ad --cellxgene-cafe --port 5005

When running from a notebook, FateAnnData keeps the convenient object method:

fadata.launch_cellxgene(use_cellxgene_cafe=True, port=5005)

Project Schedule

• Main framework code
• Document construction: instruction, tutorial, API
• Data module: FateAnnData data structure, data collection.
• Methods module: 4 backends, trajectory methods for 8 wrappers.
• Benchmark: Comprehensive metrics and benchmarking.
• Plot: Beautiful plotting capabilities.
• Downstream analysis: Interactive web platform, driver gene, and GRN analysis.

Citation

If you use Cafe in your research, please cite:

@article{huang2025cellfateexplorer,
    title={CellFateExplorer: An integrated platform for exploring cell fate},
    author={Huang, Zhaoyang and Ma, Haonan and Peng, Yuchuan and Zhao, Chenguang and Yu, Liang},
    journal={bioRxiv},
    pages={2025--02},
    year={2025},
    publisher={Cold Spring Harbor Laboratory}
}