This repository contains analysis scripts and notebooks for the TCGA BRCA dataset based on PAM50 subtypes.

HTML presentation with the results is here.

Controls:

• F to go fullscreen
• Esc to exit
• Arrows, Space, or Mouse wheel to navigate
• G then 1/{n of slide} to go to specific slide

HTSeq-FPKM-UQ data from here was used. PAM50 metadata from (Lehmann et al., 2016).

To install the required packages, run:

pip install -r requirements.txt

• results/
  • feature_importance/: Contains feature importance results.
  • diff_expression/: Contains differential expression results.
  • figures/: Contains generated figures.
  • classifiers_evaluation/: Contains classifier evaluation results.
• data/
  • processed/: Contains processed data.
  • input/: Contains raw input data.
• scripts/
  • __init__.py: Initialization script for the scripts module.
  • limma.py: Script for running the limma analysis.

• 00_preprocessing.ipynb / 00_preprocessing.py:
  • Purpose: Preprocesses the TCGA-BRCA dataset.
  • Steps:
    • Loads and merges data tables.
    • Converts Ensembl IDs to gene names.
    • Filters the dataset based on mean gene expression values (across all samples).
    • Saves the filtered dataset and metadata.
• 01_differential_expression.ipynb / 01_differential_expression.py:
  • Purpose: Performs differential expression (DE) analysis.
  • Steps:
    • Creates experiment input files for each PAM50 subtype.
    • Runs the limma analysis for each experiment using scripts/limma.py.
    • Visualizes and describes DEGs using volcano plots.
• 02_dimension_reduction.ipynb / 02_dimension_reduction.py:
  • Purpose: Reduces the dimensions of the dataset and visualizes it.
  • Steps:
    • Applies PCA, t-SNE, and UMAP for dimension reduction.
    • Visualizes the results using scatter plots.
    • Creates subsets of the dataset based on top variable genes and DEGs.
• 03_classifier.ipynb / 03_classifier.py:
  • Purpose: Implements and evaluates classifiers for breast cancer subtypes.
  • Steps:
    • Trains and evaluates multiple classifiers (e.g., GradientBoosting, SVM, Random Forest).
    • Tunes hyperparameters for top classifiers.
    • Evaluates feature importance using various methods (e.g., impurity, permutation, SHAP).

1. Differential Expression Analysis:
   • Significant differentially expressed genes (DEGs) were identified when comparing each breast cancer subtype to healthy samples, with Luminal B showing the highest number of DEGs (24.19%) and Luminal A the lowest (17.25%).
   • Significant differentially expressed genes (DEGs) were identified when comparing each breast cancer subtype to other subtypes combined, with Basal-like showing the highest number of DEGs (14.46%) and Normal-like the lowest (0.58%).
2. Dimension Reduction and Visualization:
   • PCA combined with t-SNE provided the best clustering of PAM50 subtypes, while UMAP captured global tendencies better than t-SNE alone.
3. Classifier Performance:
   • Random Forest with 100 estimators was chosen for its balance of performance and computational efficiency, achieving an accuracy of 86% and high precision and recall for most subtypes.
   • Most of key genes identified through feature importance analysis were found to be biologically relevant and associated with breast cancer prognosis and treatment response.