Related Experiment Videos
Improving hERG Cardiotoxicity Prediction via SMILES Transformer, Molecular Fingerprints, and Layer-Wise Self-Adaptive
Cheikh Beidja1, Abdelmajid Bousselham1, Zakariae En-Naimani1
12IACS Laboratory, ENSET Mohammedia, Hassan II University of Casablanca, BP 159, Bd Hassan II, Mohammedia 28810, Morocco.
Insights
A new deep learning model, TDMFLSGAT, accurately predicts drug-induced hERG channel blockade, a key factor in cardiotoxicity. This computational tool aids early drug discovery by identifying potential risks faster and more efficiently.
Area of Science:
- Computational Chemistry
- Drug Discovery
- Cardiotoxicity Prediction
Background:
- hERG channel blockade by drugs causes QT prolongation and cardiotoxicity, hindering pharmaceutical development.
- Experimental hERG liability assessment is accurate but time-consuming and costly.
- Reliable computational screening tools are needed for early-stage drug discovery.
Purpose of the Study:
- To develop a multimodal deep learning model for accurate prediction of hERG channel blockade.
- To integrate diverse molecular representations for enhanced predictive performance.
- To provide mechanistic insights into hERG liability through an interpretability framework.
Main Methods:
- Proposed TDMFLSGAT: a multimodal architecture combining Transformer (SMILES), graph attention network (structure), and molecular fingerprints (physicochemical properties).
- Integrated three complementary molecular representations into a unified model.
- Incorporated a multimodal interpretability framework for model transparency.
Main Results:
- TDMFLSGAT achieved high predictive performance: accuracy 0.823, AUC 0.901, AP 0.915.
- The model demonstrated robust performance under class imbalance, confirmed by Average Precision.
- Interpretability framework highlighted structural motifs associated with hERG blockade.
Conclusions:
- TDMFLSGAT offers accurate predictions for hERG liability, aiding early cardiotoxicity screening.
- The model provides valuable mechanistic insights, improving transparency in drug discovery.
- This tool represents a promising advancement for efficient and reliable drug development.
Abstract:
Blocking of hERG (human ether-à-go-go-related gene) potassium channels by certain drug-like molecules remains a major obstacle in pharmaceutical development, as it can induce QT interval prolongation and serious cardiotoxic effects. Experimental assessment of hERG liability is accurate but costly and time-consuming, which motivates the development of reliable computational screening tools. In this work, we propose TDMFLSGAT (transformer-enhanced deep learning model with molecular fingerprints and layer-wise self-adaptive graph attention network), a multimodal architecture that integrates three complementary molecular representations. Sequential information from SMILES strings is encoded with a Transformer, graph-based structural features are extracted using an adaptive graph attention mechanism, and physicochemical patterns are captured through a diverse set of molecular fingerprints. These modalities are fused into a unified representation for robust classification. To enhance model transparency, we further incorporate a multimodal interpretability framework that combines attention-based analyses and fingerprint-level explanations to highlight structural motifs associated with hERG blockade. Under 5-fold cross-validation, TDMFLSGAT achieves strong and well-balanced predictive performance, with an accuracy of 0.823, an AUC of 0.901, an average precision (AP) of 0.915, a sensitivity of 0.850, a specificity of 0.792, an NPV of 0.810, a PPV of 0.834, and an MCC of 0.641. The inclusion of AP confirms that these performance gains remain robust under class imbalance, providing a more realistic assessment of model reliability. Overall, these results indicate that TDMFLSGAT offers both accurate predictions and meaningful mechanistic insights, making it a promising tool for early stage cardiotoxicity screening in drug discovery.