The VSGB 2.0 model: a next generation energy model for high resolution protein structure modeling

Area of Science:

• Computational Biology
• Structural Biology
• Biophysics

Background:

• Accurate protein structure modeling is crucial for understanding biological function and developing drugs.
• Existing energy models face challenges in accurately predicting the conformations of larger protein segments, such as loops.

Purpose of the Study:

• To introduce and validate the VSGB 2.0 energy model for high-resolution protein structure prediction.
• To assess the model's performance on predicting long protein loops.

Main Methods:

• Developed the VSGB 2.0 energy model incorporating an optimized implicit solvent model and physics-based corrections.
• Trained model parameters using a crystallographic database.
• Employed advanced sampling methods and a robust protonation state assignment algorithm.
• Validated the model by predicting 115 protein loops ranging from 14 to 20 residues.

Main Results:

• VSGB 2.0 achieved high accuracy in predicting protein loop conformations.
• All lowest energy conformations had global backbone RMSDs better than 2.0 Å compared to native structures.
• Average global backbone RMSDs for loop predictions (14-20 residues) were below 1.6 Å, even for the longest loops.

Conclusions:

• The VSGB 2.0 model demonstrates significant precision and robustness in protein structure modeling.
• The model's accuracy in predicting long loops makes it suitable for real-world applications.
• Potential applications include biological function modeling and structure-based drug discovery.