Data from: Multiscale spatial atlas of human basal ganglia cell types and communities
Data files
Oct 13, 2026 version files 9.35 GB
-
Donor1_AP18_MERFSIH_adata.h5ad
1.02 GB
-
Donor1_AP18_Stereoseq_adata.h5ad
1.29 GB
-
Donor1_AP34_MERFSIH_adata.h5ad
1.52 GB
-
Donor2_AP21_Stereoseq_adata.h5ad
820.96 MB
-
Donor3_AP18_MERFSIH_adata.h5ad
588.63 MB
-
Donor3_AP19_Stereoseq_adata.h5ad
1.04 GB
-
Donor3_AP34_MERFSIH_adata.h5ad
1.01 GB
-
Donor4_AP19_MERFSIH_adata.h5ad
649.98 MB
-
Donor4_AP26_MERFSIH_adata.h5ad
1.39 GB
-
README.md
5.24 KB
Abstract
This dataset contains processed MERFISH and Stereo-seq data generated from large, intact coronal sections of human basal ganglia tissue. The MERFISH dataset includes gene expression matrices; cell-type annotations at five hierarchical resolutions—neighborhood, class, subclass, group, and cluster—consistently annotated according to the Human and Mammalian Brain Atlas (HMBA) basal ganglia consensus taxonomy; spatial modules identified through cellular community analysis; annotations of matrix and striosome compartments; cell spatial coordinates; and UMAP embeddings. The Stereo-seq dataset includes gene expression matrices and spatial module annotations derived from unsupervised clustering of spatially weighted gene expression profiles. These processed datasets support the analyses and figures presented in the associated manuscript.
This dataset is associated with the manuscript Berackey et al., Multiscale spatial atlas of human basal ganglia cell types and communities, Cell (2026),https://doi.org/10.1016/j.cell.2026.09.027
It contains the processed MERFISH and stereoseq datasets required to reproduce the analysis and figures presented in the manuscript.
There are six MERFISH datasets:
- Donor1_AP18_MERFSIH_adata.h5ad
- Donor1_AP34_MERFSIH_adata.h5ad
- Donor3_AP18_MERFSIH_adata.h5ad
- Donor3_AP34_MERFSIH_adata.h5ad
- Donor4_AP19_MERFSIH_adata.h5ad
- Donor4_AP26_MERFSIH_adata.h5ad
corresponding to the six samples used for MERFISH experiment. Similarly, there are three stereoseq datasets:
- Donor1_AP18_Stereoseq_adata.h5ad
- Donor3_AP19_Stereoseq_adata.h5ad
- Donor2_AP21_Stereoseq_adata.h5ad
associated with the three samples used for Stereo-seq experiment.
The fields in the MERFISH .h5ad files are as follows.
obs
- AP_level: Anterior-Posterior level of the section
- Donor_local_ID: Local identifier for the donor from which the section was collected from
- Donor_NIMP_NHASHID: Donor identifier for sampled registered in NIMP portal
- n_genes_by_counts: number of unique genes detected within a cell
- total_counts: Total count of transcripts
- log1p_n_genes_by_counts: log1p transformation of unique gene count within a cell
- log1p_total_counts:log1p transformation of total transcript count
- volm: volume of cell segmentation mask. Total number of pixels in 3D cellpose segmentation mask.
- leiden: ledien cluster label
- Neighborhood: cell type annotation at neighborhood level in the consensus basal ganglia cell type taxonomy
- Neighborhood_label: unique taxonomy label at neighborhood level
- Neighborhood_bootstrapping_probability: confidence of label assignment at neighborhood level
- Class: cell type annotation at class level in the consensus basal ganglia cell type taxonomy
- Class_label: unique taxonomy label at class level
- Class_bootstrapping_probability: confidence of label assignment at class level
- Subclass: cell type annotation at subclass level in the consensus basal ganglia cell type taxonomy
- Subclass_label:unique taxonomy label at subclass level
- Subclass_bootstrapping_probability: confidence of label assignment at subclass level
- Group: cell type annotation at group level in the consensus basal ganglia cell type taxonomy
- Group_label:unique taxonomy label at group level
- Group_bootstrapping_probability: confidence of label assignment at group level
- Cluster: cell type annotation at cluster level in the consensus basal ganglia cell type taxonomy
- Cluster_label:unique taxonomy label at cluster level
- Cluster_alias:u nique identifier for cluster level taxonomy node
- Cluster_bootstrapping_probability:confidence of label assignment at cluster level
- GlobalSpatialModule_annotation: Spatial modules annotation obtained from cellular community analysis
- MatrixStriosome_annotation: Matrix-Striosome annotation obtained from cellular compostion analysis of spiny projection neurons(SPN).
var
- Ensemble_ID: Ensemble ID of genes
- n_cells_by_counts: number of cells a gene is measured in.
- mean_counts: mean expression over all cells.
- log1p_mean_counts: log1p transformation of mean_counts
- pct_dropout_by_counts: percentage of cells a gene does not appear in
- total_counts: sum of cell counts for a gene
- log1p_total_counts:log1p transformation of total_counts
obsm
- X_raw: raw expression counts
- X_spatial: spatial coordinates of cells
- X_umap: UMAP embedding coordinates
uns
- Contains clustering, color assignments, UMAP, and neighbor graph, including:
'leiden', 'leiden_colors', 'log1p', 'neighbors', 'umap'
obsp
- connectivities and distances contain the cell-cell neighbor graph and distance matrix of nearset neighbors.
The fields in the Stereoseq .h5ad files are as follows.
obs
- Donor_local_ID: Local identifier for the donor from which the section was collected from
- Donor_NIMP_NHASHID:Donor identifier for sampled registered in NIMP portal
- AP_level:Anterior-Posterior level of the section
- NGenes: Number of unique genes detected within a cell
- TotalUMI:Total count of transcripts
- area: cell size
- dnbCount: total number of DNA nanoballs within the cell segmentation mask
- hemo: hemoglobu gene percentage
- mito: mitochondrial gene percentage
- ribo: ribosomal gene percentage
- SpatialModule_annotation:
var
- Gene: gene name
obsm
- spatial: spatial coordinates of cells
layers
- log1p_count: log1p transformed expression count
- raw_count: raw expression count
Code and Software
The code used for data analysis and visualization is available at the GitHub repository:
https://github.com/BBerackey/BICAN_human_basal_ganglia_spatial_transcriptomics_manuscript_2025
The dataset was generated from postmortem human basal ganglia tissue collected from four adult donors through the University of California, Irvine (UCI) Center for Neural Circuit Mapping BICAN brain bank under approval of the Institutional Review Board (IRB).
