Can a physics-based, all-atom potential find a protein's native structure among misfolded structures? I. Large scale...

Area of Science:

• Computational Biology
• Structural Bioinformatics
• Molecular Modeling

Background:

• Physics-based, all-atom energy functions (AMBER, CHARMM, OPLS-AA) and local minimization effectively discriminate between native and decoy protein structures during scoring.
• However, their application in refining protein models from initial conformations to native-like states has seen limited success.

Purpose of the Study:

• To investigate the discrepancy between scoring and refinement capabilities of physics-based potentials, specifically the AMBER potential.
• To identify reasons why the AMBER potential may not reliably guide protein models towards their native structures during conformational search.

Main Methods:

• Conformational search using molecular dynamics driven by the AMBER potential.
• Analysis of 150 nonhomologous proteins and their associated decoys to evaluate energy landscapes.

Main Results:

• For a significant number of proteins, the native state did not correspond to the lowest free energy state when using the AMBER potential for refinement.
• This indicates a limitation in the AMBER potential's ability to accurately represent the native state as the global minimum.

Conclusions:

• The AMBER potential, while effective for scoring, faces challenges in accurately refining protein models to native-like structures.
• Potential corrections to the AMBER potential function are proposed to improve its suitability for protein model refinement and achieve better native-likeness.