Proteome-wide, Structure-Based Prediction of Protein-Protein Interactions/New Molecular Interactions Viewer

Area of Science:

• Plant molecular biology
• Computational biology
• Proteomics

Background:

• Understanding the Arabidopsis (Arabidopsis thaliana) protein-protein interaction network is crucial for deciphering proteome organization.
• Previous studies have mapped only a portion of the estimated 300,000 interactions in Arabidopsis.

Purpose of the Study:

• To evaluate the efficacy of a docking algorithm in identifying genuine protein-protein interactions using predicted protein structures.
• To expand the known Arabidopsis interactome using computational structure-based predictions.

Main Methods:

• Generated 0.91 million potential interactions from 1,346 predicted Arabidopsis protein structure models.
• Ranked interactions using a docking algorithm.
• Validated top-ranked interactions through cosubcellular enrichment analysis and yeast two-hybrid assays.

Main Results:

• A significant enrichment of real interactions was observed among the top-ranked predictions.
• The structure-based computational predictions achieved 63% of the success rate of published interactions in yeast two-hybrid tests.
• The method outperformed random protein pair selection by 2.7 times.

Conclusions:

• Protein structure information and docking algorithms offer a valuable approach for interactome exploration and network reconstruction.
• The study provides a new perspective on mapping protein-protein interactions in Arabidopsis.
• Predicted interactions and associated tools are made publicly accessible to facilitate further research.