# Tooling attribution and licenses

This `tooling/` tree is included for **transparency and reproducibility** of how
the TCGA-NSCLC diagnostic WSIs in `wsi/TCGA-*` were obtained. It is **not** a
substitute for the GDC Data Transfer Tool or for TCGA/GDC data-use terms.

## 1. Download scripts / manifests (third-party)

**Source repository (quote / cite this):**

- **Name:** TCGA-lung-histology-download  
- **Author:** GeorgeBatch  
- **URL:** https://github.com/GeorgeBatch/TCGA-lung-histology-download  
- **Description (upstream):** “Downloading and Creating labels for the TCGA lung Whole Slide Images”  
- **Snapshot:** see `tcga-lung-histology-download/UPSTREAM_COMMIT.txt` (commit hash pinned at mirror time)

The files under `tooling/tcga-lung-histology-download/` are a mirror of that
repository (excluding `.git`), including:

- `0-download-LUSC-and-LUAD.sh` — wraps `gdc-client download` for LUAD + LUSC  
- `tcga-download/gdc_manifest.2023-10-03-TCGA-*.txt` — GDC manifests used for this cohort  
- `README.md` and helper scripts / notebooks from upstream  

### License status of the upstream code

As of mirror time, the GitHub repository
[GeorgeBatch/TCGA-lung-histology-download](https://github.com/GeorgeBatch/TCGA-lung-histology-download)
lists **`license: null`** (no `LICENSE` file, no SPDX license on the repo metadata).
We redistribute this snapshot **only for scientific transparency / provenance**
of our download workflow. **All rights in that code remain with the upstream
author(s).** If you reuse or redistribute the mirrored scripts beyond viewing them
here, check the upstream repository for an updated license or obtain permission
from the author.

This Hub dataset does **not** claim ownership of GeorgeBatch’s scripts.

## 2. GDC client (not redistributed here)

Downloads were performed with the NCI GDC Data Transfer Tool (`gdc-client`),
from:

- https://gdc.cancer.gov/access-data/gdc-data-transfer-tool  

We do **not** redistribute the GDC client binary in this dataset. On our
download host we used a thin local wrapper (`gdc-client-fork`) only to force
Python 3.14’s multiprocessing start method to `fork` (upstream `gdc-client`
pickle issue). That wrapper is environment-specific and is **not** required to
interpret the data on Hub; see `thesis-tcga-download-notes/download_TCGA_NSCLC.md`.

## 3. TCGA / GDC data (the WSIs themselves)

The slide files under `wsi/TCGA-*` are **not** licensed by GeorgeBatch’s repo.
They come from:

- [The Cancer Genome Atlas (TCGA)](https://www.cancer.gov/tcga)  
- via [NCI Genomic Data Commons (GDC)](https://gdc.cancer.gov/)  

Terms: NIH Genomic Data Sharing (GDS) unrestricted-access / GDC open-access
rules for these diagnostic slides — **no re-identification**, cite TCGA/GDC.
See the dataset card (`README.md`) gated acknowledgment and License section.

Acknowledgment:

> The results shown here are in whole or part based upon data generated by the
> TCGA Research Network: https://www.cancer.gov/tcga. Data were obtained from
> the NCI Genomic Data Commons (GDC): https://gdc.cancer.gov/.

## 4. Thesis-specific notes

`tooling/thesis-tcga-download-notes/` contains our runbook for the full
TCGA-NSCLC pull on the thesis compute host (paths, concurrency, md5 verify).
Those notes are thesis documentation; the **authoritative download recipes and
manifests** remain the upstream GeorgeBatch repo cited above plus GDC.

## How to re-download from scratch (high level)

1. Install `gdc-client` from NCI GDC.  
2. Clone or use `tooling/tcga-lung-histology-download/` from this dataset.  
3. Follow upstream `README.md`, using the `2023-10-03` manifests.  
4. Verify `.svs` md5 against the manifests (see also `manifests/` in this dataset).