---
license: mit
language:
- en
tags:
- spatial-transcriptomics
- 10x
- homo
pretty_name: Human DLPFC Visium · 4 adjacent slices
size_categories:
- 10K<n<100K
---

        # Human DLPFC Visium · 4 adjacent slices

        Curated, ready-to-load spatial transcriptomics dataset.

        ## Source

        - Paper: [Maynard et al., Nat. Neurosci. 2021](https://www.nature.com/articles/s41593-020-00787-0)
        - Canonical download: spatialLIBD R package · slices 151673–151676

        ## Scale

        | Property | Value |
        |---|---|
        | Technology | 10x Genomics Visium |
        | Species | Homo sapiens |
        | Tissue | Human dorsolateral prefrontal cortex |
        | Sections / slices | 4 |
        | Total cells / spots | 14,243 |

        ## Files

        - `h5ad/adata_1.h5ad`
- `h5ad/adata_2.h5ad`
- `h5ad/adata_3.h5ad`
- `h5ad/adata_4.h5ad`

        Each `.h5ad` follows the AnnData spec:
        - `.X` — gene expression matrix (cells × genes), sparse where natural
        - `.obs` — per-cell annotations (see "Metadata" below)
        - `.obsm['spatial']` — `(n_cells, 2)` float32 spatial coordinates
        - (where present) `.layers['count']` — raw integer counts
        - (where present) `.obsm['spatial3d']` — `(n_cells, 3)` float32 (x, y, z=section)

        ## Metadata (`obs` columns)

        `layer`, `in_tissue`, `array_row`, `array_col`

        ## Notes

        4 of the 12-slice Maynard atlas (slices 151673–151676), preprocessed as ready-to-load AnnData. `obs['layer']` carries the manual L1–L6 + WM annotation.

        ## Usage

        ```python
import scanpy as sc
from huggingface_hub import snapshot_download
d = snapshot_download(repo_id='Shaow/dlpfc_visium_4slice', repo_type='dataset')
adata_st_list = [sc.read_h5ad(f'{d}/h5ad/adata_{i}.h5ad') for i in range(1, 5)]
```


        ## Citation

        If you use this dataset, please cite the source paper above.

        ## License

        MIT for the curation/preparation. Underlying data inherits the license of
        the upstream publication (typically CC-BY-4.0); please see the source paper.