A Tutorial for Drug Response Prediction and Graph Neural Network at APBJC 2024

Contact: Yurui Chen, [email protected]

This repository provides a tutorial on building a Graph Neural Network (GNN) for drug response prediction from scratch. The project includes data processing, model layers, and a Jupyter Notebook to run the full pipeline.

• Drug response prediction plays a crucial role in precision medicine, aiming to determine how various drugs will impact specific cell lines or patients, ultimately assisting in creating personalized treatment strategies.

• The GNN approach is highly effective for learning representations from small molecule graphs, making it well-suited for this type of task.

We will cover data processing, building the model, and running predictions step by step. You are expected to:

1. Get familiar with the drug response prediction problem.
2. Gain basic knowledge of the Graph Neural Network (GCN and GIN).
3. Understand the pipeline to train and evaluate machine learning and deep learning models for drug responses.

We will use the GDSC project as the drug response dataset.

For workshop participants:

Join the WIFI:

SSID: Nahart Free WI-FI

Password: nahart78

The github repo is: https://github.com/matcyr/APBJC2024_GNN_Tutorial short link to the repo:

bit.ly/3AaFxDO

We will use colab to run the tutorial. Please click on

or scan:

For experienced python users:

To start, in your terminal, run:

git clone https://github.com/matcyr/APBJC2024_GNN_Tutorial.git

Then go to the project directory, by:

cd APBJC2024_GNN_Tutorial

Setting Up the Environment

The tutorial is based on Python. The key dependencies include:

• pandas
• numpy
• torch
• torch_geometric

Note: The provided dependencies are configured for CPU usage only. If you want to train the model using a GPU, please refer to the PyTorch website and PyTorch Geometric website for instructions on setting up a GPU-backed environment.

1. Conda + Linux

If you have a linux platform with conda installed, you can simply install the environment from the provided environment.yml file. Run the following commands in your terminal:

conda env create -f environment.yml -n <env_name>
conda activate <env_name>

2. Pip

If you do not have Conda installed, or if you are using a Mac/Windows laptop. You can set up the environment using pip by following these instructions:

Step 1: Create a Virtual Environment

Run the following command in your terminal to create a virtual environment:

python -m venv <env_name>

or:

python3 -m venv <env_name>

Step 2: Activate the Virtual Environment

• macOS/Linux: Run the following command in your terminal:

  source <env_name>/bin/activate

• Windows: Run the following command in your Command Prompt (cmd):

  <env_name>\Scripts\activate
  

Step 3: Install Dependencies

After activating the virtual environment, run the following command to install the required dependencies:

sh install.sh

This will create a virtual environment named <env_name> and install the required dependencies using the install.sh script.

The Tutorial Notebook

Run the following command to open the tutorial notebook:

jupyter notebook notebooks/APBJC_tutorial.ipynb