HiGraphDTI: Hierarchical Graph Representation Learning for Drug-Target Interaction Prediction

TL;DR

HiGraphDTI tackles DTI prediction by introducing a hierarchical drug representation that captures atoms, motifs, and global structure, coupled with a broadened target representation through attentional feature fusion. A hierarchical attention mechanism then models cross-level drug–target interactions and provides interpretable insights into binding mechanisms. Empirical results across four benchmarks show consistent improvements over six state-of-the-art methods, with ablation and attention-visualization analyses validating the contribution and interpretability of the approach. The work advances DTI discovery by delivering both higher predictive performance and actionable biological interpretations from multi-scale molecular graphs.

Abstract

The discovery of drug-target interactions (DTIs) plays a crucial role in pharmaceutical development. The deep learning model achieves more accurate results in DTI prediction due to its ability to extract robust and expressive features from drug and target chemical structures. However, existing deep learning methods typically generate drug features via aggregating molecular atom representations, ignoring the chemical properties carried by motifs, i.e., substructures of the molecular graph. The atom-drug double-level molecular representation learning can not fully exploit structure information and fails to interpret the DTI mechanism from the motif perspective. In addition, sequential model-based target feature extraction either fuses limited contextual information or requires expensive computational resources. To tackle the above issues, we propose a hierarchical graph representation learning-based DTI prediction method (HiGraphDTI). Specifically, HiGraphDTI learns hierarchical drug representations from triple-level molecular graphs to thoroughly exploit chemical information embedded in atoms, motifs, and molecules. Then, an attentional feature fusion module incorporates information from different receptive fields to extract expressive target features.Last, the hierarchical attention mechanism identifies crucial molecular segments, which offers complementary views for interpreting interaction mechanisms. The experiment results not only demonstrate the superiority of HiGraphDTI to the state-of-the-art methods, but also confirm the practical ability of our model in interaction interpretation and new DTI discovery.

Figures (7)

• Figure 1: The overview architecture of HiGraphDTI.
• Figure 2: Hierarchical graph representation construction. In the diagram, solid lines represent bidirectional edges, and dashed lines represent unidirectional edges.
• Figure 3: The overview architecture of feature fusion module for protein. The high-level feature is mapped to the same dimension as the low-level using transposed convolution and then input into the AFF module for fusion. Taking the result as high-level features, repeat the operation.
• Figure 4: The architecture of AFF module, which utilizes operation $M$ to compute the attention matrix for weighted aggregation of inputs $\mathbf{I}_1$ and $\mathbf{I}_2$.
• Figure 5: Ablation experiment results on the GPCR dataset