Modeling Protein-Protein Complexes by Combining pyDock and AlphaFold.

Area of Science:

• Structural Biology
• Computational Biology
• Artificial Intelligence in Biochemistry

Background:

• Experimental structures for most protein-protein complexes are unavailable.
• Computational docking and AI-based methods like AlphaFold are key strategies for structural modeling.
• AlphaFold excels at individual protein structure prediction but faces challenges with complexes.

Purpose of the Study:

• To present a practical method for combining AlphaFold and pyDock.
• To improve the accuracy of structural predictions for challenging protein-protein complexes.
• To leverage AI and classical methods for enhanced structural modeling.

Main Methods:

• Utilizing AlphaFold for generating diverse structural models of protein complexes.
• Employing pyDock's energy-based scoring function to refine and select accurate models.
• Integrating AI-driven predictions with physics-based scoring for validation.

Main Results:

• Demonstrated a practical workflow for integrating AlphaFold and pyDock.
• Showcased improved structural prediction capabilities for difficult protein-protein interactions.
• Validated the combined approach for identifying correct complex models.

Conclusions:

• Combining AlphaFold with pyDock offers a powerful strategy for accurate protein complex structure prediction.
• This integrated approach addresses limitations of standalone methods for complex cases.
• The method provides a valuable tool for advancing research in structural biology and drug discovery.