A Multi-Objective Approach for Protein Structure Prediction Based on an Energy Model and Backbone Angle Preferences

Area of Science:

• Computational biology
• Biophysics
• Structural bioinformatics

Background:

• Protein structure prediction (PSP) aims to determine a protein's 3D structure from its amino acid sequence.
• Existing optimization methods based on energy models require further improvement for accuracy and structural similarity.
• Understanding protein folding mechanisms is crucial for advancing PSP.

Purpose of the Study:

• To introduce a novel backbone angle preference factor to enhance ab initio protein structure prediction.
• To develop and evaluate a multiobjective optimization approach integrating energy models and backbone angle preferences.
• To improve the accuracy and similarity of predicted protein structures.

Main Methods:

• Developed a multiobjective optimization approach for ab initio PSP.
• Incorporated a novel backbone angle preference factor alongside traditional energy models.
• Validated the approach using 75 diverse amino acid sequences from the CB513 dataset (22-88 amino acids).

Main Results:

• The multiobjective optimization approach demonstrated superior performance compared to typical energy models.
• Achieved lower root-mean-square deviation (RMSD), indicating higher accuracy in structure prediction.
• The method proved effective on a highly dissimilar and meaningful benchmark dataset.

Conclusions:

• The proposed multiobjective optimization approach, incorporating backbone angle preferences, significantly facilitates ab initio protein structure prediction.
• This novel method offers improved accuracy and structural similarity over existing energy-based models.
• The findings contribute to advancing computational methods for understanding protein folding and structure.