A spatial transcriptomics atlas of live donors reveals unique zonation patterns in the healthy human liver

Sample Group Description

The dataset includes the following human & non-human samples:

Human samples:

-Live healthy donors (LHD; n = 8), labeled as M

-Patients with liver pathology (adjacent normal tissue sampled; n = 8), labeled as P

Non-human samples:

-Wild boar (n = 2), labeled as non_human_P

-Cow (n = 2), labeled as non_human_C

-Domesticated pig (n = 3), labeled as non_human_PD

Visium:

LHDs and adjacent normal samples:

Human loupe files:

This folder includes the Loupe Browser-compatible files (16 total) corresponding to the human samples (M1–M8, P2, P3, P6, P7, P14, P17, P18, P21) for downstream visualization and exploration.

Human_h5_files

This folder contains .h5 formatted output files from Space Ranger for the 16 human samples.

Human_Spatial_transcriptomics_data:

This folder contains spatial transcriptomics data from 16 human samples.
For each sample the following files are included:

counts_ALL.csv – full gene expression matrix

counts_UTT.csv – filtered matrix (UTT: under the tissue)

tissue_positions_list.csv – spatial barcode coordinates

scalefactors_json.json – image scaling information

tissue_hires_image.png – high-resolution histology image

Non Human Loupe files:

This folder contains Loupe Browser-compatible files for the 7 non-human samples (C1, C2, P1, P2, PD1, PD2, PD3).

Non_human_h5_files

This folder contains.h5 formatted output files from Space Ranger for the 7 non-human samples

Non_Human_Spatial_transcriptomics_data:

This folder includes spatial transcriptomics data from 7 non-human samples (C1, C2, P1,P2, PD1, PD2, PD3).

counts_ALL.csv – full gene expression matrix

counts_UTT.csv – filtered matrix (UTT: under the tissue)

tissue_positions_list.csv – spatial barcode coordinates

scalefactors_json.json – image scaling information

tissue_hires_image.png – high-resolution histology image

VisumHD:

This folder contains spatial transcriptomics data from 10x Genomics Visium HD for human liver samples:

M1_VisiumHD.cloupe – Loupe Browser visualization file for patient M1, showing spatial transcriptomics data at 8 μm bin resolution.

M2_VisiumHD.cloupe – Loupe Browser visualization file for patient M2, showing spatial transcriptomics data at 8 μm bin resolution.

M6_VisiumHD.cloupe – Loupe Browser file for a Visium HD slide that includes two tissue sections. The tissue at the bottom of the slide corresponds to patient M6, which is the one analyzed in the downstream dataset (marked under ‘patients’ as ‘M6-high quality’). Data is shown at 8 μm bin resolution.

visiumHD_data_M2_M6.h5ad – A filtered and integrated .h5ad file containing single-cell–resolved spatial gene expression data from both M2 and M6 samples.

-Resolution: Cells. Based on single-cell segmentation (see “Liver Cell Atlas using Visium HD” method).

-Cell filtering: Only cells detected via segmentation and that have passed quality filters were included.

-UMI threshold: Cells with fewer than 200 UMIs were excluded.

-Batch correction: Harmony was applied to correct sample-specific effects prior to UMAP visualization.

-Format: .h5ad (AnnData format, compatible with Scanpy).

-Includes: Single-cell expression matrix, spatial coordinates, Harmony-corrected UMAP, cluster identity, and metadata.

M6:

This folder contains spatial transcriptomics data (8*8 μm) for sample M6, generated using the 10x Genomics Visium HD platform.

NOTES:

-The gene expression matrices (*.h5) come from the full slide output of Space Ranger, including both tissue sections (like the M6 Loupe file ).

-The spatial metadata files (*.json, *.tif, .csv) refer to the cropped region, corresponding to the bottom tissue, which is the actual M6 sample used in downstream analysis.

-This is the raw Space Ranger output, prior to cell segmentation or high-level filtering (apart from the default filtered feature matrix).

-This data reflects raw 8 μm resolution bins, not single-cell segmentations.

-For downstream analysis based on cell segmentation, refer to the visiumHD_data_M2_M6.h5ad file in the top-level VisiumHD folder.

Gene Expression Matrices (uncropped – both tissues included):

filtered_feature_bc_matrix_8um.h5
raw_feature_bc_matrix_8um.h5

Spatial Metadata (cropped – M6 tissue only):

scalefactors_json.json
Images: tissue_hires_image.tif / tissue_lowres_image.tif / tissue_fullres_image.tif
tissue_positions.csv  - Barcode-to-position table corresponding only to the cropped region, i.e., the M6 tissue. Only the barcodes listed in this file are relevant to M6 and should be used to extract or analyze this tissue’s expression data from the full matrix.

M2:

This folder contains the full, unmodified output of the 10x Genomics Visium HD Space Ranger for the M2 liver tissue sample (8X8 um resolution).

filtered_feature_bc_matrix_8um.h5 
raw_feature_bc_matrix_8um.h5
scalefactors_json.json
Images: tissue_hires_image.tif / tissue_lowres_image.tif
tissue_positions_orig.csv 

M1:

This folder contains the full, unmodified output of the 10x Genomics Visium HD Space Ranger for the M1 liver tissue sample (8X8 um resolution)

filtered_feature_bc_matrix_8um.h5 
raw_feature_bc_matrix_8um.h5
scalefactors_json.json
Images: tissue_hires_image.tif / tissue_lowres_image.tif
tissue_positions_orig.csv 

snRNAseq:

This folder contains single-nucleus RNA-seq (snRNA-seq) data from four human liver samples (M5, M6, M7, M8). Data was generated using Cell Ranger multi.

single_nuc_RNAseq.cloupe - Output from Cell Ranger multi. data from all four samples.

snRNAseq.h5ad -   Processed and filtered .h5ad file containing single-nucleus expression data from M5–M8, integrated into one dataset.

-Filtering includes standard QC (e.g., low-gene/UMI exclusion, mitochondrial content, etc.)

-Batch correction: Harmony was applied to correct sample-specific effects prior to UMAP visualization.

-Format: .h5ad (AnnData format, compatible with Scanpy).

-Includes: expression matrix, spatial coordinates, Harmony-corrected UMAP, cluster identity and metadata.

M5, M6, M7, M8

Each sample folder contains raw and filtered matrices generated by Cell Ranger:

-sample_filtered_feature_bc_matrix

-sample_raw_feature_bc_matrix

MERFISH:

For both samples- M5 and M8, each sample folder contains:

-cell_by_gene.csv

-cell_metadata.csv

-detected_transcripts.csv

As an additional validation step, we performed highly multiplexed immunofluorescence (PhenoCycler™) using a panel of 48 antibodies targeting a broad range of parenchymal and non-parenchymal cell types in 2 live-donor livers (LHDs) and 4 adjacent-normal liver samples (Supplementary Table 5 in the linked dataset). This data can be found at:

https://zenodo.org/records/17730820