A predicted structural interactome reveals binding interference from intrinsically disordered regions

Area of Science:

• Structural biology
• Computational biology
• Genomics

Background:

• Protein-protein interactions (PPIs) are crucial for cellular functions, forming complex networks.
• High-throughput methods advance PPI understanding, but molecular details, especially in non-mammalian species like Drosophila, remain poorly characterized.
• Deep learning advances enable network-level prediction of molecular details in cellular pathways.

Purpose of the Study:

• To predict and examine protein-protein interactions in Drosophila using AlphaFold2 multimer.
• To assess the contribution of physical and functional datasets to prediction accuracy.
• To investigate the role of intrinsically disordered regions in high-confidence PPI predictions.

Main Methods:

• Utilized AlphaFold2 multimer for predicting protein-protein interactions in Drosophila.
• Integrated both physical and functional interaction datasets.
• Performed detailed structural analysis of predicted complexes, focusing on intrinsically disordered regions.

Main Results:

• Functional associations significantly improved the confidence of predicted protein-protein interactions.
• Intrinsically disordered regions were identified as important in high-confidence predicted interactions.
• An interactive web interface was developed to visualize and explore the predicted Drosophila interactome.

Conclusions:

• Incorporating functional interaction data enhances the accuracy of predicting physical protein-protein interactions.
• Intrinsically disordered regions play a significant role in mediating protein-protein interactions.
• The developed web interface facilitates further research into Drosophila protein interactions and hypothesis generation.