Datasets:
metadata
license: mit
tags:
- spatial-transcriptomics
- gene-regulatory-network
- xenium
- spacetravlr
- mouse
- skin
pretty_name: SpaceTravLR outputs
size_categories:
- 100B<n<1T
SpaceTravLR dataset hub
Precomputed SpaceTravLR outputs: per-gene beta matrices (*_betadata.feather), run metadata, and optional per-sample .h5ad exports.
Layout
spacetravlr/
├── tonsil/ # placeholder / demo gene outputs
└── xenium_skin_mixed/
├── run.toml # shared training config for this cohort
├── manifest.json # sample index and upload metadata
├── sample12/
├── sample13/
├── sample14/
├── sample15/
└── sample16/
Each xenium_skin_mixed/sample*/ directory contains:
| File / folder | Description |
|---|---|
*_betadata.feather |
Per-modulator beta matrices from SpaceTravLR training |
sample*.h5ad |
Condition-filtered AnnData export for the sample |
sample*.csv |
Summary table for the run |
condition_label.txt |
Condition identifier |
spacetravlr_run_repro.toml |
Reproducibility config (paths rewritten on download) |
spacetravlr_run_summary.html |
HTML training summary |
log/ |
Training logs |
Not uploaded: perturbations/ folders (in-silico perturbation grids).
Cohort: xenium_skin_mixed
Mouse skin Xenium data with mixed samples. Training used xenium_skin_mixed.h5ad with sample_id as the condition column and cell_type_int as the cluster annotation.
| Sample | Approx. size | Beta files |
|---|---|---|
| sample12 | 59 GB | 2,809 |
| sample13 | 103 GB | 2,807 |
| sample14 | 66 GB | 2,810 |
| sample15 | 100 GB | 2,801 |
| sample16 | 89 GB | 2,801 |
Download
# Entire xenium_skin_mixed cohort
hf download Koushul/spacetravlr xenium_skin_mixed --repo-type dataset --local-dir ./spacetravlr
# Single sample
hf download Koushul/spacetravlr xenium_skin_mixed/sample12 --repo-type dataset --local-dir ./sample12
Usage with SpaceTravLR
Point spatial_viewer or spacetravlr-perturb at a downloaded sample directory containing *_betadata.feather and the matching spacetravlr_run_repro.toml.
Citation
If you use these outputs, please cite SpaceTravLR and the underlying Xenium skin dataset.