---
license: mit
task_categories:
  - other
tags:
  - spatial-transcriptomics
  - spatial-proteomics
  - 3d-reconstruction
  - image-registration
pretty_name: Space-map example datasets
---

# Space-map example datasets

Serial-section single-cell coordinate datasets used with
[Space-map](https://github.com/a12910/space-map) for atlas-level 3D
reconstruction. Each file is a gzipped CSV of cell coordinates with a
per-section layer label.

| File | Platform / tissue | Layer column | Coord columns | Layers |
|------|-------------------|--------------|---------------|--------|
| `xenium_colon.csv.gz` | Xenium (transcriptomics), colon | `layer` (1–20) | `x`, `y` | 20 |
| `codex_colon.csv.gz` | CODEX (proteomics), colon | `array` (S1–S16) | `x`, `y` | 16 |
| `codex_duodenum_raw.csv.gz` | CODEX (proteomics), duodenum — raw | `index` (S1–S16) | `raw_x`, `raw_y` | 16 |
| `codex_duodenum_aligned.csv.gz` | CODEX duodenum — reference aligned | `layer` (S1–S16) | `x`, `y` | 16 |

Notes:

- `xenium_colon.csv.gz` layer labels were renumbered to a contiguous 1–20 range.
- The two `codex_duodenum_*` files share row order and layer labels: `raw` holds
  the input coordinates, `aligned` holds a reference alignment for comparison.
- Additional columns (e.g. `cell_type`, `cell_id`) are preserved where present.

## Usage

```python
from huggingface_hub import hf_hub_download
import pandas as pd

path = hf_hub_download("a12910/spacemap-data", "xenium_colon.csv.gz",
                       repo_type="dataset")
df = pd.read_csv(path)
```

See the Space-map repository for registration examples
(`examples/run_platforms.py`).