Sequence-Based Protein-Protein Interaction Prediction and Its Applications in Drug Discovery

Area of Science:

• Computational biology
• Bioinformatics
• Drug discovery

Background:

• Aberrant protein-protein interactions (PPIs) are implicated in numerous human diseases.
• Targeting these harmful interactions presents a promising therapeutic strategy.
• Computational methods, particularly deep learning, are advancing PPI prediction.

Purpose of the Study:

• To review state-of-the-art sequence-based PPI prediction methods.
• To explore the impact of PPI prediction on target identification and drug discovery.
• To emphasize rigorous model assessment in PPI prediction.

Main Methods:

• Overview of data sources and curation techniques for PPI prediction.
• Survey of traditional similarity-based and deep learning-based PPI predictors.
• Emphasis on transformer architectures in deep learning for PPI prediction.

Main Results:

• Sequence-based PPI prediction offers a viable alternative to structure-based methods.
• Examples of PPI prediction in proteomics, target identification, and therapeutic design are provided.
• The review underscores the importance of data quality and model assessment.

Conclusions:

• Sequence-based PPI prediction is a broadly applicable tool in biomedical research.
• These methods significantly contribute to identifying therapeutic targets and designing novel drugs.
• Advancements in computational approaches are revolutionizing disease treatment strategies.