GraphDeep-hERG: Graph Neural Network PharmacoAnalytics for Assessing hERG-Related Cardiotoxicity

Pharm Res. 2025 Apr;42(4):579-591. doi: 10.1007/s11095-025-03848-w. Epub 2025 Mar 26.

Abstract

Purpose: The human Ether-a-go-go Related-Gene (hERG) encodes rectifying potassium channels that play a significant role during action potential repolarization of cardiomyocytes. Blockade of the hERG channel by off-target drugs can lead to long QT syndrome, significantly increasing the risk of proarrhythmic cardiotoxicity. Traditional hERG screening methods are effort-demanding and time-consuming. Thus, it is essential to develop computational methods to utilize the existing knowledge for faster and more accurate in silico screening. Although with wide use of deep learning/machine learning algorithms, existing computational models often rely on manually defined atomic features to represent atom nodes, which may overlook critical underlying information. Thus, we want to provide a new method to learn the atom representation automatically.

Methods: We first developed an automated atom embedding model using deep neural networks (DNNs), trained with 118,312 compounds collected from the ZINC database. We then trained a Graph neural networks (GNNs) model with 7909 ChEMBL compounds as the classifying part. The integration of our atom embedding model and GNN models formed a classifier that could effectively distinguish between hERG inhibitors and non-inhibitors.

Results: Our atom embedding model achieved 0.93 accuracy in representing structures. Our best GNN model achieved an accuracy of 0.84 and outcompeted traditional machine-learning models, as well as published AI-driven models, in external testing.

Conclusions: These results highlight the potential of our automated atom embedding model as a standard for generating robust molecular representations. Its integration with advanced GNN algorithms offers promising assistance for screening hERG inhibitors and accelerating drug discovery and repurposing.

Keywords: HERG; alzheimer’s disease; atom-type embedding model; cardiovascular toxicity; deep learning/machine learning; graph neural networks.

MeSH terms

  • Action Potentials / drug effects
  • Algorithms
  • Cardiotoxicity* / etiology
  • Computer Simulation
  • Deep Learning
  • ERG1 Potassium Channel* / antagonists & inhibitors
  • Ether-A-Go-Go Potassium Channels* / antagonists & inhibitors
  • Graph Neural Networks
  • Humans
  • Machine Learning
  • Neural Networks, Computer*
  • Potassium Channel Blockers* / adverse effects
  • Potassium Channel Blockers* / chemistry
  • Potassium Channel Blockers* / pharmacology

Substances

  • Potassium Channel Blockers
  • Ether-A-Go-Go Potassium Channels
  • ERG1 Potassium Channel
  • KCNH2 protein, human