Protein-Protein Docking Using EMAP in CHARMM and Support Vector Machine: Application to Ab/Ag Complexes.

Area of Science:

• Computational Biology
• Structural Bioinformatics
• Molecular Modeling

Background:

• Accurate modeling of antibody-antigen (Ab/Ag) complexes is crucial for understanding immune responses and developing therapeutics.
• Existing computational methods often face challenges in generating and identifying native Ab/Ag conformations efficiently.

Purpose of the Study:

• To assess the efficacy of the EMAP method within the CHARMM program for generating correct Ab/Ag complex structures.
• To develop a support vector machine (SVM) classifier for distinguishing native Ab/Ag conformations from decoys using binding free energy components.

Main Methods:

• Evaluation of the EMAP (Effective Model Potential) method for Ab/Ag complex structure generation.
• Development and training of an SVM classifier using features derived from a thermodynamic cycle of binding free energy.
• Testing on 24 Ab/Ag complexes from the protein-protein docking benchmark v3.0, assessed by CAPRI criteria.

Main Results:

• The EMAP method successfully generated medium-quality native conformations for all tested Ab/Ag complexes.
• The SVM classifier ranked medium/high-quality native conformations within the top six predictions out of thousands.
• Demonstrated feasibility of Ab-Ag docking using diverse protein representations (grid-based, united-atom, all-atom) within CHARMM.

Conclusions:

• The EMAP method provides a reliable approach for generating near-native Ab/Ag complex conformations.
• The developed SVM classifier effectively enhances the accuracy of identifying correct Ab/Ag binding poses.
• This work facilitates flexible and accurate computational antibody-antigen docking strategies.