Shaow/mousehippocampus_starmap_zeng

    # Mouse Hippocampus STARmap PLUS · 8-month TauPS2APP rep1+rep2

    Curated, ready-to-load spatial transcriptomics dataset.

    ## Source

    - Paper: [Zeng et al., Nat. Neurosci. 2023](https://www.nature.com/articles/s41593-022-01251-x)
    - Canonical download: Zenodo doi:10.5281/zenodo.7683099 (CSVs assembled into AnnData)

    ## Scale

    | Property | Value |
    |---|---|
    | Technology | STARmap PLUS (2,766-gene panel) |
    | Species | Mus musculus |
    | Tissue | Mouse hippocampus (8-month TauPS2APP AD model) |
    | Sections / slices | 2 |
    | Total cells / spots | 16,388 |

    ## Files

    - `h5ad/adata_8m_disease_rep1.h5ad`

• h5ad/adata_8m_disease_rep2.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)
  
    `top_level_cell_type`, `sub_level_cell_type`, `region`, `region_merged`
  
    ## Notes
  
    Two STARmap PLUS replicates of an 8-month TauPS2APP mouse hippocampus (rep1: 8,186 cells; rep2: 8,202 cells; both × 2,766 genes). Assembled from the Zenodo CSVs (`expression_matrix_raw.csv` + `metadata.csv` + `spatial_8months-disease-replicate_{1,2}.csv`) into ready-to-load AnnData.
  
    ## Usage
  
    ```python
  

import scanpy as sc from huggingface_hub import snapshot_download d = snapshot_download(repo_id='Shaow/mousehippocampus_starmap_zeng', repo_type='dataset') a1 = sc.read_h5ad(f'{d}/h5ad/adata_8m_disease_rep1.h5ad') a2 = sc.read_h5ad(f'{d}/h5ad/adata_8m_disease_rep2.h5ad')

        ## 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.