Clustering estrogen receptor-positive breast cancer tumors based on hormonal response type

Project in Machine Learning (CS-433)

EPFL, 2019

Abstract

We employ unsupervised machine learning techniques to cluster subtypes of estrogen receptor-positive breast cancer, which is the most common variant worldwide. Clustering is done according to hormone responses obtained from in vivo models of patient-derived xenografts. Our results facilitate more targeted treatment of patients, responding to the urgent need for personalized medicine to treat breast cancer.

Architecture

results/ directory is for output and plots
scripts/ directory contains all the code
constants.py includes constants used throughout the project
helpers.py includes various helper functions
load.py includes functions for loading and manipulating the data
plots.py includes plotting functions
clustering_helpers.py includes helper functions used for the cluster analysis
data_analysis.ipynb is the Jupyter Notebook file that includes the exploratory data analysis and visualizations
cluster_analysis.ipynb is the Jupyter Notebook file that includes the cluster analysis

Dependencies

Data handling and ML libraries:
- sklearn (v. 0.21.3)
- pandas (v. 0.25.1)
- numpy (v. 1.16.5)
Plotting libraries:

Data

Download here

Reproduction

Clone or fork the repository
Download the data and add the data/ folder to the root of the project
Install Jupyter Notebook
Install the abovementioned libraries
Run data_analysis.ipynb to reproduce the data analysis results
Run cluster_analysis.ipynb to reproduce the cluster analysis results

Authors

Lisa Dratva, lisa.dratva@epfl.ch
Michal Pleskowicz, michal.pleskowicz@epfl.ch
Valentin Oliver Loftsson, valentin.loftsson@epfl.ch

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

We thank Fabio De Martino, our supervisor at the BRISKEN lab, for his constant guidance and support throughout the learning process.

valentinoli/pdx