HBGSA: Hydrogen Bond Graph with Self-Attention for Drug-Target Binding Affinity Prediction

TL;DR

HBGSA is a novel graph neural network model that encodes hydrogen bond spatial features with self-attention and correlation loss to improve drug-target binding affinity prediction accuracy.

Contribution

The paper introduces HBGSA, a new model that effectively incorporates hydrogen bond spatial information and correlation-aware loss for better affinity prediction.

Findings

HBGSA outperforms baseline methods on benchmark datasets.

Hydrogen bond modeling improves prediction accuracy.

Pearson correlation loss enhances model generalization.

Abstract

Accurate prediction of drug-target binding affinity accelerates drug discovery by prioritizing compounds for experimental validation. Current methods face three limitations: sequence-based approaches discard spatial geometric constraints, structure-based methods fail to exploit hydrogen bond features, and conventional loss functions neglect prediction-target correlation, a key factor for identifying high-affinity compounds in virtual screening. We developed HBGSA (Hydrogen Bond Graph with Self-Attention), a 3.06M-parameter model that encodes hydrogen bond spatial features. HBGSA uses graph neural networks to model hydrogen bond spatial topology with self-attention enhancement and Pearson correlation loss. Experimental results on PDBbind Core Set and CSAR-HiQ dataset demonstrate that HBGSA outperforms baseline methods with strong generalization capability. Ablation studies confirm the effectiveness of hydrogen bond modeling and Pearson correlation loss.