Protein side-chain modeling with a protein-dependent optimized rotamer library

Area of Science:

• Biochemistry
• Structural Biology
• Computational Biology

Background:

• Predicting protein side chain conformations is a critical but challenging problem in structural biology.
• Existing methods often rely on rotamer libraries derived from statistical analysis of known protein structures.

Purpose of the Study:

• To develop a novel, statistics-free method for predicting protein side chain conformations.
• To improve the accuracy and diversity of predicted side chain packings.

Main Methods:

• Developed a protein-dependent rotamer library (PDRL) based solely on stereochemical considerations.
• Optimized the PDRL for each protein individually to restore conformational diversity.
• Integrated PDRL with self-consistent mean field (SCMF) sampling and a physics-based scoring function into the SCMF-PDRL method.

Main Results:

• The SCMF-PDRL method demonstrated favorable performance compared to established methods like SCAP, OPUS-Rota, and SCWRL4.
• Tested on large datasets (831 and 378 proteins), the method showed competitive accuracy for energy-minimized structures.

Conclusions:

• The SCMF-PDRL method offers a promising alternative for accurate protein side chain conformation prediction.
• The protein-dependent rotamer library approach enhances the prediction of diverse and native-like conformations.