TCGA_Data_Download_Guide.md

TCGA Data Download Guide

This guide explains how to download RNA-Seq and clinical data from The Cancer Genome Atlas (TCGA) for use with OncoLearn.

Overview

OncoLearn provides convenient download scripts for fetching TCGA data from the UCSC Xena Browser. The scripts download:

• RNA-Seq data: STAR FPKM-UQ normalized gene expression data
• Clinical data: Patient phenotype and clinical information

Quick Start

Download All Cohorts

To download all available cancer cohorts at once:

bash ./scripts/data/download_all_tcga.sh

This will download and extract all cohort data to data/GDCdata/.

Available Cancer Cohorts

The following TCGA cancer cohorts are supported:

Cohort Code	Cancer Type	Script
TCGA-BRCA	Breast Invasive Carcinoma	download_tcga_brca.sh
TCGA-COAD	Colon Adenocarcinoma	download_tcga_coad.sh
TCGA-LAML	Acute Myeloid Leukemia	download_tcga_laml.sh
TCGA-LUAD	Lung Adenocarcinoma	download_tcga_luad.sh
TCGA-LUSC	Lung Squamous Cell Carcinoma	download_tcga_lusc.sh
TCGA-MESO	Mesothelioma	download_tcga_meso.sh
TCGA-SKCM	Skin Cutaneous Melanoma	download_tcga_skcm.sh

Download Individual Cohorts

To download specific cancer cohorts, use the individual scripts:

# Lung Adenocarcinoma
bash ./scripts/data/download_tcga_luad.sh

# Lung Squamous Cell Carcinoma
bash ./scripts/data/download_tcga_lusc.sh

# Breast Cancer
bash ./scripts/data/download_tcga_brca.sh

# Colon Cancer
bash ./scripts/data/download_tcga_coad.sh

# Melanoma
bash ./scripts/data/download_tcga_skcm.sh

# Mesothelioma
bash ./scripts/data/download_tcga_meso.sh

# Acute Myeloid Leukemia
bash ./scripts/data/download_tcga_laml.sh

Data Size Considerations

Each cohort varies in size:

• Small cohorts (e.g., LAML): ~50-100 MB
• Medium cohorts (e.g., COAD, LUSC): 100-300 MB
• Large cohorts (e.g., BRCA, LUAD): 300-600 MB

Total download size for all cohorts: Approximately 2-3 GB

Ensure you have sufficient disk space before downloading.

Data Storage Structure

All downloaded data is organized in the data/GDCdata/ directory:

data/
└── GDCdata/
    ├── TCGA-BRCA.clinical.tsv
    ├── TCGA-BRCA.star_fpkm-uq.tsv
    ├── TCGA-COAD.clinical.tsv
    ├── TCGA-COAD.star_fpkm-uq.tsv
    ├── TCGA-LAML.clinical.tsv
    ├── TCGA-LAML.star_fpkm-uq.tsv
    └── ...

Each cohort has two files:

• TCGA-{COHORT}.clinical.tsv - Clinical and phenotype data
• TCGA-{COHORT}.star_fpkm-uq.tsv - Gene expression data

Data Processing

After downloading, you can use the preprocessing notebooks to merge and prepare the data:

1. Merge clinical and expression data:

   • Notebook: notebooks/data/preprocess_merge_cohort_data.ipynb
   • Output: Merged files in data/processed/

2. Prepare multimodal data:

   • Notebook: notebooks/data/preprocess_multimodal_data_fusion.ipynb

3. Exploratory data analysis:

   • Notebook: notebooks/data/eda_cohorts.ipynb