Chemi-net: a graph convolutional network for accurate drug property prediction

TL;DR

Chemi-Net is a novel graph convolutional neural network that predicts drug ADME properties more accurately than traditional methods, potentially speeding up drug discovery processes.

Contribution

The paper introduces Chemi-Net, a data-driven deep learning model that outperforms existing machine learning methods in ADME property prediction without relying on domain-specific features.

Findings

Chemi-Net significantly outperforms Cubist in ADME prediction accuracy.

Deep neural networks can replace traditional domain-specific descriptors in drug property prediction.

Enhanced prediction accuracy may accelerate drug discovery processes.

Abstract

Absorption, distribution, metabolism, and excretion (ADME) studies are critical for drug discovery. Conventionally, these tasks, together with other chemical property predictions, rely on domain-specific feature descriptors, or fingerprints. Following the recent success of neural networks, we developed Chemi-Net, a completely data-driven, domain knowledge-free, deep learning method for ADME property prediction. To compare the relative performance of Chemi-Net with Cubist, one of the popular machine learning programs used by Amgen, a large-scale ADME property prediction study was performed on-site at Amgen. The results showed that our deep neural network method improved current methods by a large margin. We foresee that the significantly increased accuracy of ADME prediction seen with Chemi-Net over Cubist will greatly accelerate drug discovery.