Cross-Modal Interaction-Aware Progressive Fusion Network for Drug-Target Interaction Prediction

Area of Science:

• Computational chemistry
• Bioinformatics
• Machine learning

Background:

• Drug-target interaction (DTI) prediction is crucial for accelerating drug discovery.
• Deep learning models have advanced DTI prediction, but challenges remain in capturing and fusing bidirectional cross-modal information.
• Existing methods struggle with effectively integrating diverse data modalities for DTI analysis.

Purpose of the Study:

• To propose a novel deep learning fusion framework, the cross-modal interaction-aware progressive fusion network (CIPFN), for improved DTI prediction.
• To address the limitations of existing methods in capturing and fusing bidirectional DTI information.
• To enhance the accuracy and efficiency of identifying potential drug-target relationships.

Main Methods:

• Developed a deep learning fusion framework named CIPFN.
• Introduced a bidirectional interaction-aware module for fine-grained drug-protein interaction alignment.
• Implemented a progressive fusion network with gated and convolutional blocks for critical information extraction.

Main Results:

• CIPFN demonstrated significant improvements across five benchmark datasets.
• Achieved superior performance in AUROC, AUPRC, F1, sensitivity, and accuracy compared to state-of-the-art methods.
• Effectively captured and fused cross-modal information for enhanced DTI prediction.

Conclusions:

• The proposed CIPFN framework offers a powerful approach for DTI prediction.
• CIPFN effectively addresses the challenges of bidirectional information fusion in DTI analysis.
• This method holds promise for accelerating the drug discovery pipeline through improved DTI identification.