Predicting Protein Dimer Structures Using MELD × MD.

Area of Science:

• Computational Biology
• Structural Biology
• Biophysics

Background:

• Protein-protein interactions are crucial for cellular functions.
• Accurately predicting protein complex structures remains a significant challenge in structural biology.
• Existing methods often oversimplify protein flexibility or cannot reliably rank different conformations.

Purpose of the Study:

• To develop and validate a computational approach for predicting native protein-protein docked conformations.
• To assess the efficacy of combining protein contact prediction with atomistic molecular dynamics simulations.
• To demonstrate the utility of free energy calculations in identifying correct protein dimer structures.

Main Methods:

• Utilized the ClusPro server to generate potential protein-protein contact models.
• Employed Modeling Employing Limited Data (MELD) accelerated molecular dynamics (MELD × MD) for atomistic simulations.
• Calculated free energies for various docked conformations derived from initial models.

Main Results:

• The combined ClusPro and MELD × MD approach successfully identified 16 out of 20 native dimer structures.
• Native structures were consistently ranked among those with the lowest calculated free energies.
• Atomistic molecular dynamics free energy calculations proved effective in distinguishing native from non-native complexes.

Conclusions:

• The integration of ClusPro and MELD × MD offers a robust method for predicting protein-protein interactions.
• Free energy calculations derived from accelerated molecular dynamics are a reliable metric for validating predicted protein complex structures.
• This approach advances the capability to accurately model protein dimerization relevant to biological processes.