This repository provides an open-source, reproducible workflow for analyzing 10x Xenium spatial transcriptomics data using segmentation-free spatial factorization (FICTURE / punkst), with extensions for ROI-based transcript cropping.
The project combines:
- interactive transcript-level ROI cropping using Napari
- integration with punkst / FICTURE for segmentation-free modeling
- orchestration of single-sample and multi-sample analyses
- documentation designed for reproducibility and academic reuse
This repository is intended to serve as both a research pipeline and a citable software artifact.
Xenium datasets often contain millions of transcripts spanning large tissue sections. While whole-slide analysis is valuable, many biological questions focus on specific anatomical regions or require joint analysis across multiple samples.
This workflow supports:
- region-of-interest (ROI)–focused analysis
- transcript-level filtering without cell segmentation
- consistent analysis across many Xenium samples
- scalable workflows for large spatial datasets
A high-level conceptual description is provided in
docs/overview.md.
- Interactive visualization and ROI definition using Napari
- Polygon-based transcript filtering
- Deterministic cropping given the same polygon
- Outputs compatible with downstream punkst / FICTURE analysis
See cropping/README.md for details.
- punkst / FICTURE is included as a Git submodule
- Upstream code is not copied or modified directly
- Authorship, licensing, and versioning are preserved
Original repository: https://github.com/yichen-si/punkst
This repository contributes workflows and orchestration around FICTURE, not a reimplementation of the method itself.
- Single-sample workflows
- Multi-sample joint analysis workflows
- SVG flowcharts suitable for presentations and posters
See workflows/README.md.
Reproducibility considerations are documented in
docs/reproducibility.md, including:
- deterministic vs interactive steps
- ROI provenance
- submodule versioning
- environment considerations
Zenodo is used to archive tagged releases and provide a DOI.
If you use this workflow, please cite:
Sabir Y, Huang M, Cruz-Martín A.
Xenium ROI Cropping + FICTURE Workflows.
Zenodo. https://doi.org/10.5281/zenodo.18490063
Please also cite the original FICTURE publication when using downstream factorization.
This project is released under the MIT License.
See the LICENSE file for details.
- Large Xenium datasets are not included in this repository.
- This workflow is designed for research and exploratory analysis.
- Contributions and issues are welcome via GitHub.