Predicting the protein interaction landscape of a free-living bacterium with pooled-AlphaFold3

Area of Science:

• Computational Biology
• Genomics
• Structural Biology

Background:

• Protein-protein interactions (PPIs) are crucial for cellular functions.
• Predicting PPIs genome-wide is computationally intensive and has been limited.
• Existing methods struggle with scalability for large-scale genomic analysis.

Purpose of the Study:

• To introduce a novel, scalable method for genome-wide protein-protein interaction (PPI) prediction.
• To overcome the computational and utility limitations of applying structure prediction programs like AlphaFold3 to entire genomes.
• To generate a comprehensive dataset of pairwise PPIs for Mycoplasma genitalium.

Main Methods:

• Developed pooled-protein-protein interaction (PPI) prediction, a technique for genome-scale PPI screens.
• Applied pooled-PPI prediction to predict all pairwise PPIs in Mycoplasma genitalium.
• Utilized AlphaFold3 for structure prediction, significantly reducing computational jobs and inference time compared to paired approaches.

Main Results:

• Predicted 113,050 pairwise PPIs in Mycoplasma genitalium using only 2,027 AlphaFold3 jobs.
• Achieved a ~2-fold reduction in inference time and a ~100-fold reduction in computational jobs.
• The generated dataset was highly predictive of known interactions and revealed a widespread size bias in AlphaFold interface scores.
• Identified protein-protein interfaces in macromolecular complexes and uncovered novel biological insights in M. genitalium.

Conclusions:

• Pooled-protein-protein interaction (PPI) prediction is a highly scalable and accurate method for functional genomics.
• This technique dramatically improves the efficiency of genome-scale PPI screens.
• The study establishes a powerful new tool for uncovering protein-protein interactions and advancing biological discovery.