ExPDRUG Pipeline

This repository contains the ExPDRUG pipeline, designed for drug discovery using gene expression data. Below you will find instructions on how to set up and run the pipeline.

File Structure

util/logger.py

• Contains logging functions and result output features.

util/config.py

• Handles file input/output paths and hyperparameter adjustments.

util/data_processor.py

• Manages data processing for model training, including:
  • Creating and managing masking matrices between layers.
  • Shuffling functionality for permutation tests.

util/model.py

• Defines the neural network model and relevance score computation logic.
• Includes the implementation of the custom loss function.

util/trainer.py

• Handles model training, k-fold validation, and relevance score computation.
• Implements permutation test functionality for model validation.

main.py

• The main script to run the entire pipeline.
• Orchestrates data loading, model training, and interpretation using LRP, IG, or GSEA methods.

1. Data Preparation

Data Sources

• Hetionet: Hetionet GitHub
• Reactome: Reactome Data Download

Gene Expression Data

• COVID-19: COVID-19 Data
• COVID-19 Severity Score Information: Severity Score Information
• GBM: GBM Data
• Alzheimer's: Alzheimer's Data

Ensure that the raw gene expression files for experiments are placed in the data folder.

2. Data Filtering for ExPNet Training

Run the scripts in the data_processor folder to filter the data for training ExPNet. Refer to the pipeline in that folder for specific instructions.

3. Model Training, Relevance Score Computation, and Permutation Test

1. Set the file paths and hyperparameters in the .util/config.py file.
2. Execute the main.py script.

python main.py

4. Drug Discovery

Run the scripts in the RWR folder to perform drug discovery.

Installation

Install the required packages using the requirements.txt file:

pip install -r requirements.txt