ml-bio-atg

A Comparative Study of Machine Learning and Graph Neural Network Models for Predicting Bioactivity of Environmental Chemicals

This repository contains the code and Jupyter notebooks used in the study titled "A Comparative Study of Machine Learning and Graph Neural Network Models for Predicting Bioactivity of Environmental Chemicals" by Matthew Adams, Grace Patlewicz, and Imran Shah.

Authors

• Matthew Adams (ORAU, Oak Ridge, TN, 37830, USA)
• Grace Patlewicz (Center for Computational Toxicology & Exposure (CCTE), US Environmental Protection Agency, Research Triangle Park, Durham, NC, 27709, USA)
• Imran Shah (Center for Computational Toxicology & Exposure (CCTE), US Environmental Protection Agency, Research Triangle Park, Durham, NC, 27709, USA)

Corresponding author:
Imran Shah
Address: Center for Computational Toxicology & Exposure (CCTE), US EPA, 109 TW Alexander Drive, RTP, NC 27711, USA
Tel: +1 919-541-1391
Email: [email protected]

Overview

Repository Contents

• data/: Contains datasets used in the study.
• notebooks/: Jupyter notebooks with the implementation of machine learning and graph neural network models.
• src/: Source code for data processing, model training, and evaluation.
• README.md: This README file.
• env.yml: Conda environemnt

Requirements

The requirements can be installed by creating a conda environment:

conda env create -f env.yml