HMBVIP: A Novel Hierarchical Multi-Bio-View Intelligent Prediction Networks for Drug-Target Interaction Prediction.

Area of Science:

• Computational Biology
• Bioinformatics
• Drug Discovery

Background:

• Drug-target interaction (DTI) prediction is vital for accelerating drug discovery.
• Multiview learning models integrate diverse biological data for improved DTI prediction accuracy and robustness.
• Current methods struggle with single-scale tokenizers and shallow data integration, missing biological granularity and hierarchy.

Purpose of the Study:

• To address limitations in current DTI prediction models.
• To develop a multiscale biological tokenizer and a hierarchical multi-bio-view learning (HMBV) approach.
• To enhance the accuracy and interpretability of DTI predictions.

Main Methods:

• Proposed a "bio-token" concept and a multiscale biological tokenizer for capturing features at varying resolutions.
• Developed a hierarchical multi-bio-view learning (HMBV) approach implemented in the HMBVIP network.
• Utilized end-to-end network architecture for DTI prediction.

Main Results:

• The HMBVIP network demonstrated superior performance on benchmark datasets compared to existing state-of-the-art models.
• Hierarchical multiview fusion enriched representations with multidimensional biological context.
• The approach enhanced both prediction accuracy and biological interpretability.

Conclusions:

• The proposed HMBV approach and HMBVIP network offer a significant advancement in DTI prediction.
• Multiscale feature extraction and hierarchical data integration are key to improving DTI prediction.
• HMBVIP provides a more accurate and interpretable framework for drug discovery.