A distance-dependent atomic knowledge-based potential and force for discrimination of native structures from decoys

Area of Science:

• Computational biology
• Structural bioinformatics
• Biophysics

Background:

• Accurately determining protein structure is crucial for understanding biological function.
• Distinguishing native protein folds from misfolded decoys remains a significant challenge in structural biology.

Purpose of the Study:

• To present a novel computational model for discriminating native protein structures from designed decoy structures.
• To utilize mechanical interatomic forces derived from statistical contact preferences for structure evaluation.

Main Methods:

• Modeling proteins as systems of springs to calculate interatomic forces.
• Developing a potential function based on statistical contact preferences in known protein structures.
• Defining a scoring function based on the calculated forces for each atom within a 3D structure.

Main Results:

• The model assigns higher scores to atoms experiencing lower forces, indicating more stable configurations.
• The optimal protein structure is identified as the one with the highest overall score.
• The model's performance was evaluated using several decoy sets, demonstrating its potential in structure discrimination.

Conclusions:

• The proposed model offers a novel approach to protein structure discrimination using mechanical principles.
• This method provides a promising tool for identifying correctly folded proteins among potential decoys.
• Further application and refinement of this force-based scoring function could enhance protein structure prediction and analysis.