ESCC spatial-to-bulk validation framework

This repository contains the analysis code and reproducibility materials for the manuscript:

A tiered spatial-to-bulk reproducibility framework supports an association-level CAF/ECM stromal-remodeling phenotype in esophageal squamous cell carcinoma

The study is a public-data bioinformatics workflow. It applies a tiered validation framework that connects:

1. spatially nominated hypotheses,
2. bulk transcriptomic validation, and
3. source-table checks from published supplementary quantitative tables.

The repository is intended to support code availability and reproducibility for the public-data analyses. It does not contain controlled-access data, newly generated human-subject data, manuscript drafts, administrative submission files, full-text PDFs, or temporary local files.

Repository contents

• scripts/: Python scripts used for the public-data workflow, independent patient/source-table checks, supplemental transferability analysis, and S1 workbook assembly from generated TSV outputs.
• project_config.yaml: project-level configuration and declared study scope.
• requirements.txt: Python dependencies used by the submitted code package.
• manifests/: public-data source manifest.
• supporting_information/: submitted processed tables copied from the manuscript package.
• reproducibility_check.tsv: local check summary.
• S2_Code_manifest.json: manifest for the submitted code package.
• .zenodo.json and CITATION.cff: metadata used for DOI archiving.

Data sources

The workflow uses public resources only, including TCGA/GDC/UCSC Xena, GEO cohorts GSE47404 and GSE53625, HRA003627 source data, HRA008846 supplementary tables, and GDSC2. Dataset accessions, URLs, dates, checksums where available, and processed-output paths are listed in the manifest files and in the manuscript supporting information.

Reproducibility scope

This repository is designed to reproduce the reported public-data analyses and submitted tables from the available public data layers. Some source tables are redistributed as processed supporting information in the manuscript package; raw third-party datasets remain available from their original repositories. The workflow does not perform causal inference, wet-lab validation, prospective clinical validation, or expression-drug response modeling.

Quick start

Use Python 3.12 or a compatible Python 3 environment.

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

On Windows PowerShell:

py -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install -r requirements.txt

Run scripts from the repository root after reviewing project_config.yaml and the source manifest. Output paths are configured inside the scripts and may need to be adjusted if the repository is moved.

Citation

Cite the Zenodo metadata repair public code release for the manuscript-submission version:

https://github.com/afa-cloud/escc-spatial-bulk-validation-framework-code

Release tag: v1.0.13-submission-zenodo-fix.

The GitHub release page records the exact release commit.

Zenodo concept DOI for this repository: https://doi.org/10.5281/zenodo.19826728.

The code-availability statement for peer review should cite the archived release tag, its Zenodo version DOI, and the submitted S2 Code package.

License

Code in this repository is released under the MIT License. Third-party datasets remain governed by the terms of their original repositories and publications.