naono-git/mgcnn_alkaloid

Molecular Graph Convolutional Neural Networks

mgcnn_alkaloid: Molecular Graph Convolutional Neural Networks for Prediction of Alkaloid Biosynthesis Bathways

Description

Source code for our paper "Classification of alkaloids into starting substances on biosynthetic pathway using graph convolutional neural networks".

Requirements

Python >= 3.6.6 NumPy >= 1.15.4 Pandas >= 0.22.0 TensorFlow 1.6.0 DeepChem 2.1.0

Files

• mgcnn_alkaloid.py: Training and evaluation of Molecular Graph Convolutional Neural Networks for alkaloid biosynthesis pathway prediction.
• data/alkaloid_data.csv: Molecular data of alkaloids.

Usage

1. Five fold cross validation

python alkaloid_data.csv

2. Verify one segment of five fold cross validation

python alkaloid_data.csv [0-4]

Licence

MIT

Author

Naoaki ONO

naono-git

mailto:[email protected]

Reference

[Currently under review]