HiF-DTA: Hierarchical Feature Learning Network for Drug-Target Affinity Prediction

2025-10-31

AI Summary

Existing sequence-based drug–target affinity (DTA) prediction methods struggle to jointly model global semantics and local topology, while drug representations are confined to flat sequences, lacking atomic-, substructural-, and molecular-level multi-scale features. To address these limitations, we propose a dual-path hierarchical network: one path employs an enhanced sequence encoder to jointly capture global semantics and local structural patterns of both proteins and drugs; the other path—novelly introduced on sequence inputs—constructs a drug multi-scale representation hierarchy and incorporates multi-scale bilinear attention for cross-level fusion. The model integrates graph attention, sequence embeddings, and dual-path feature interaction. Extensive experiments demonstrate significant improvements over state-of-the-art methods on Davis, KIBA, and Metz benchmarks. Ablation studies validate the effectiveness of joint global–local modeling and multi-scale feature fusion.

Technology Category

• Machine Learning: Deep Neural Architectures and Foundation Models
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
• Data Mining & Knowledge Management: Scalability, Parallel & Distributed Systems

Application Category

• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding

Abstract

Accurate prediction of Drug-Target Affinity (DTA) is crucial for reducing experimental costs and accelerating early screening in computational drug discovery. While sequence-based deep learning methods avoid reliance on costly 3D structures, they still overlook simultaneous modeling of global sequence semantic features and local topological structural features within drugs and proteins, and represent drugs as flat sequences without atomic-level, substructural-level, and molecular-level multi-scale features. We propose HiF-DTA, a hierarchical network that adopts a dual-pathway strategy to extract both global sequence semantic and local topological features from drug and protein sequences, and models drugs multi-scale to learn atomic, substructural, and molecular representations fused via a multi-scale bilinear attention module. Experiments on Davis, KIBA, and Metz datasets show HiF-DTA outperforms state-of-the-art baselines, with ablations confirming the importance of global-local extraction and multi-scale fusion.

Problem

• Predicting drug-target affinity using sequence-based deep learning
• Modeling global sequence semantics and local topological features simultaneously
• Representing drugs with multi-scale atomic, substructural, and molecular features

Innovation

• Hierarchical network with dual-pathway feature extraction
• Multi-scale drug modeling via bilinear attention fusion
• Global-local sequence semantic and topological integration