Spatial multi-omics and deep learning reveal fingerprints of immunotherapy response and resistance in hepatocellular carcinoma - Processed datasets and notebook

This entry accompanies the manuscript "Spatial multi-omics and deep learning reveal fingerprints of immunotherapy response and resistance in hepatocellular carcinoma" (preprint).

Contents:

1. Processed CODEX MIF Data (CODEX-MIF.zip)

CODEX data from 48 fields of view (identified by region_id) are provided in the following files:

• {region_id}.cell_data.csv: A table with three columns—CELL_ID, X, and Y—containing the centroid coordinates of all segmented cells in the corresponding field of view.
• {region_id}.cell_types.csv: A table with two columns—CELL_ID and CELL_TYPE—containing cell type annotations for all cells in the region.
• {region_id}.cell_features.csv: A table with multiple columns—CELL_ID, SIZE, ORIENTATION, etc.—containing morphological features of all cells in the corresponding field of view.
• {region_id}.expression.csv: A table with multiple columns—CELL_ID, DAPI, CD45, etc.—providing normalized protein biomarker expression levels for all cells in the region.

2. Processed Visium ST Data (Visium-ST.zip)

Visium spatial transcriptomics (ST) data from 22 samples are provided in the following files:

• visium_all.h5ad: An AnnData object containing raw gene expression counts for all spatial transcriptomics spots across the 22 samples.
• {region_label}_tissue_positions_list.csv: A table with multiple columns—barcode, transformed coords x, transformed coords y, etc.—providing ST spot coordinates co-registered with the corresponding CODEX MIF data. The transformed coordinates align with the cell centroid coordinates specified in the CODEX datasets.

3. Metadata (metadata.csv)

A metadata file in CSV format that maps region_id to key attributes, including:

• Region label
• Patient ID
• Pre-/Post-treatment status
• Response to immunotherapy
• Treatment type

4. Data Analysis Notebook (notebook_and_resources.zip)

A zip file containing:

• A Jupyter Notebook to reproduce the analytical workflows for generating the main figure panels in the manuscript.
• Intermediate data files related to plotting and visualization.

Citation:

When using this dataset, please cite the original manuscript (DOI: 10.1101/2025.06.11.656869) and this Zenodo entry (DOI:  10.5281/zenodo.15392699).