A survey on deep learning for drug-target binding prediction: models, benchmarks, evaluation, and case studies

Area of Science:

• Computational Biology
• Drug Discovery
• Artificial Intelligence

Background:

• Conventional drug discovery is costly, slow, and often fails.
• Artificial intelligence (AI) offers advanced solutions for complex biological problems.
• Drug-target binding (DTB) is crucial for effective drug discovery.

Purpose of the Study:

• To review recent deep learning models for drug-target binding (DTB) prediction.
• To analyze the evolution of AI methodologies in DTB prediction.
• To identify challenges and future directions in AI-driven drug discovery.

Main Methods:

• Analysis of deep learning models, benchmark datasets, and metrics for DTB prediction.
• Examination of AI methodologies: heterogeneous networks, graph-based, attention-based, and multimodal approaches.
• Case studies using compound libraries against cancer-related protein targets.

Main Results:

• Deep learning has revolutionized drug discovery, shifting methodologies from network-based to advanced AI architectures.
• Demonstrated the utility of AI in predicting drug-target interactions and affinities.
• Identified limitations of current DTB prediction models.

Conclusions:

• Deep learning provides a quantitative framework for understanding drug-target relationships.
• AI significantly speeds up the identification of novel drug candidates.
• Future research should focus on refining AI models for more accurate DTB prediction.