Lung Adenocarcinoma Survival Analysis and Biomarker Discovery Pipeline

This repository contains an end-to-end Python pipeline for survival analysis and exploratory biomarker discovery in lung adenocarcinoma using a real public cancer cohort.

The workflow uses the LUAD TCGA PanCancer Atlas study from cBioPortal, combines clinical survival data with RNA-seq expression data, screens candidate biomarkers for association with overall survival, constructs a simple expression-based risk score, and generates figures, summary tables, and a report automatically.

Dataset

The analysis uses the public Lung Adenocarcinoma PanCancer Atlas study from cBioPortal:

• Study identifier: luad_tcga_pan_can_atlas_2018
• Sources:
  • https://datahub.assets.cbioportal.org/luad_tcga_pan_can_atlas_2018.tar.gz
  • https://download.cbioportal.org/luad_tcga_pan_can_atlas_2018.tar.gz

This study provides both patient survival data and RNA-seq expression data, making it suitable for an end-to-end biomarker discovery workflow.

Workflow

The pipeline performs the following steps:

1. downloads and extracts the LUAD study archive
2. loads patient-level clinical data, sample-level clinical data, and RNA-seq expression data
3. filters to primary tumor samples with valid overall survival information
4. tests candidate biomarkers for survival association using a log-rank framework
5. ranks biomarkers by statistical signal
6. builds a simple multi-gene risk score from the top biomarkers
7. stratifies patients into high-risk and low-risk groups
8. generates Kaplan-Meier curves, biomarker plots, and summary tables
9. writes a markdown summary report

Output Files

After a successful run, the pipeline writes:

• output/processed/top_biomarkers.tsv
• output/processed/risk_scored_patients.tsv
• output/reports/analysis_summary.md
• output/figures/kaplan_meier_risk_groups.png
• output/figures/top_biomarkers.png
• output/figures/risk_score_distribution.png
• output/figures/best_biomarker_expression.png

Repository Structure

.
├── run_pipeline.py
├── requirements.txt
├── pyproject.toml
├── src/survival_pipeline/
│   ├── analysis.py
│   ├── data.py
│   └── pipeline.py
└── tests/

Running the Pipeline

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python3 run_pipeline.py

Implementation Notes

This project is intended to demonstrate practical survival analysis and biomarker discovery using real public clinical and molecular data. The biomarker screen and risk model are exploratory rather than clinical-grade, but they show how clinical endpoints and RNA-seq measurements can be integrated into a reproducible computational biology workflow.

Testing

python3 -m unittest discover -s tests -v

License

MIT License. See LICENSE.