AlphaFold and Docking Approaches for Antibody-Antigen and Other Targets: Insights From CAPRI Rounds 47-55

Area of Science:

• Computational Biology
• Structural Biology
• Bioinformatics

Background:

• Accurate modeling of biomolecular interactions is crucial in computational biology.
• The Critical Assessment of PRedicted Interactions (CAPRI) experiment benchmarks prediction methods.
• Assessing protein-protein and protein-DNA complex structures is a key challenge.

Purpose of the Study:

• To evaluate the performance of a laboratory team in recent CAPRI rounds.
• To assess the utility of AlphaFold and other modeling tools for predicting complex structures.
• To investigate the application of AlphaFold in modeling immune recognition complexes.

Main Methods:

• Utilized ZDOCK, Rosetta, and ZRANK for modeling, refinement, and scoring.
• Employed adaptations of AlphaFold for generating structural models.
• Submitted predictions for 10 CAPRI modeling targets, including protein-protein and protein-DNA complexes.

Main Results:

• Achieved near-native models for an antibody-peptide target using AlphaFold.
• Generated a highly accurate model for an antibody-MHC complex.
• Demonstrated the effectiveness of AlphaFold-based protocols in predictive modeling.

Conclusions:

• AlphaFold-based protocols are valuable for predictive protein complex modeling, especially in immune recognition.
• Confidence metrics from AlphaFold require careful consideration during model selection.
• Continued development and application of computational tools are essential for advancing structural biology.