Protein Interaction Network Reconstruction with a Structural Gated Attention Deep Model by Incorporating Network...

Area of Science:

• Computational Biology
• Bioinformatics
• Systems Biology

Background:

• Protein-protein interactions (PPIs) are crucial for cellular functions and understanding disease.
• Existing machine learning models for PPI prediction using only protein sequences lack robustness and accuracy.

Purpose of the Study:

• To develop a novel deep learning framework, Structural Gated Attention Deep (SGAD), for improved protein-protein interaction network reconstruction (PINR).
• To enhance the robustness and prediction accuracy of PPI network analysis.

Main Methods:

• Developed the SGAD model, a deep learning framework for PINR.
• Augmented protein sequence descriptors, PPI network topological features, and information flow.
• Implemented a gating mechanism within the SGAD model to enhance noise robustness.

Main Results:

• SGAD achieved high performance on 11 independent and 1 combined dataset, with Area Under the Curve (AUC) values ranging from 0.83 to 0.93.
• The SGAD model outperformed other existing prediction models.
• An SGAD ensemble effectively captured characteristics of protein pairs via a two-layer neural network.

Conclusions:

• The SGAD model represents a significant advancement in protein-protein interaction network reconstruction.
• SGAD offers a robust and accurate tool for exploring the complex biological space of PPIs.
• This framework has the potential to deepen our understanding of biological events and disease pathogenesis.