somiya-khan01/tcga_dashboard_cancer_analysis

🧬 TCGA Breast Cancer RNA-Seq Dashboard

Interactive Streamlit dashboard for exploring DESeq2 differential expression results from TCGA-BRCA data.

⚡ Quickstart

# 1. Clone / copy your project folder
cd tcga_dashboard

# 2. Create and activate a virtual environment (recommended)
python -m venv venv
source venv/bin/activate          # Mac / Linux
# venv\Scripts\activate           # Windows

# 3. Install dependencies
pip install -r requirements.txt

# 4. Launch the dashboard
streamlit run app.py

The app opens automatically at http://localhost:8501

📁 Required Input Files

Upload these three CSV files via the sidebar in the app.

1. deseq2_results.csv — DESeq2 output

Export from R with:

write.csv(as.data.frame(res_ordered), "deseq2_results.csv", quote = FALSE)

Expected columns (DESeq2 standard output):

The first column becomes gene after import. Make sure row names are gene IDs.

2. normalized_counts.csv — Normalized count matrix

Export from R with:

norm_counts <- counts(dds, normalized = TRUE)
write.csv(norm_counts, "normalized_counts.csv", quote = FALSE)

Expected format:

• —Rows = genes (row names are gene IDs)
• —Columns = sample IDs (must match metadata index)

3. sample_metadata.csv — Sample condition labels

Export from R with:

write.csv(sample_info, "sample_metadata.csv", quote = FALSE)

Expected format:

• —Row names = sample IDs (must match counts matrix column names)
• —`condition` column = "Tumor" or "Normal"

🖥️ Dashboard Sections

🛠️ Dependencies

🔧 Troubleshooting

Heatmap shows no genes → Gene IDs in DESeq2 results and counts matrix must match exactly (both Ensembl IDs, or both gene symbols).

PCA shows "No matching samples" → Column names in normalized_counts.csv must exactly match row names in sample_metadata.csv.

Slow performance on large counts matrices → Pre-filter to the top 5,000 most variable genes before export, or subsample columns to the 100 samples analyzed.

Dashboard built with Streamlit · Plotly · scikit-learn