Key result
A hERG-tuned transfer learning GNN outperforms other models predicting in vivo cardiotoxicity, identifying ~25,000 cardiotoxic chemicals.
Why the study?
Evaluating cardiotoxicity in vivo is labor-intensive and time-consuming, while in silico models have been limited in prediction accuracy by small-scale in vivo toxicity datasets.
A transfer learning-based graph neural network model leveraging in vitro hERG binding data improves the prediction of in vivo cardiotoxicity for large-scale chemical screening.
Should not yet change practice; leaves open human relevance of large-scale in silico cardiotoxicity predictions from animal data.
Evaluating chemical hazards, such as cardiotoxicity, is critical for sound chemical management but is challenging due to the labor-intensive and time-consuming nature of in vivo testing. In silico models can efficiently and cost-effectively screen potential cardiotoxic compounds from a large number of chemicals. However, the available small-scale in vivo toxicity data sets prevent models from achieving superior prediction accuracy. Herein, a transfer learning-based graph neural network (TL-GNN) model for predicting in vivo cardiotoxicity was developed by fine-tuning from a pretrained model on a data set regarding the human ether-a-go-go-related gene binding, a molecular initiating event that may ultimately cause cardiotoxicity. The TL-GNN model not only outperformed the GNN without TL and classic machine learning models, but also those fine-tuned from GNN models pretrained on a nonmechanistic octanol–water partition coefficient data set and Bidirectional Encoder Representations from Transformers models pretrained on a data set with over a million unlabeled compounds. A structure–activity landscape-based method was adopted to characterize the applicability domains of the TL-GNN model. The TL-GNN model with defined applicability domains was applied to screen approximately 600,000 chemicals across diverse categories, identifying over 25,000 as cardiotoxic. The robustness of the TL-GNN model suggests the potential of the modeling strategy for complex in vivo toxicity by learning from adverse outcome pathway (AOP)-related in vitro targets.
No takes yet. Share an insight, caveat, or question.
Zhang et al. (2026) studied Cardiotoxicity (n=600,000). Transfer learning-based graph neural network (TL-GNN) model vs. GNN without TL and classic machine learning models was evaluated on Prediction accuracy for in vivo cardiotoxicity. A transfer learning-based graph neural network model fine-tuned on hERG binding data outperformed other models in predicting in vivo cardiotoxicity, identifying over 25,000 cardiotoxic chemicals.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: