Protein three-dimensional structure generation with an empirical hydrophobic penalty function

Area of Science:

• Computational biology
• Structural bioinformatics
• Biophysics

Background:

• Accurate protein three-dimensional structure prediction is a significant challenge in computational biology.
• Developing accurate residue-level functions is essential for advancing protein folding simulations.
• Existing methods require improved accuracy in distinguishing native protein conformations.

Purpose of the Study:

• To evaluate an empirical hydrophobic penalty function for its ability to approximate protein folds.
• To assess the predictive power of this function in distinguishing native protein structures from incorrect ones.
• To investigate the limitations of the penalty function in global protein structure prediction.

Main Methods:

• Generated protein conformations by randomly altering main chain dihedral angles.
• Applied an empirical hydrophobic penalty function based on residue neighborhood.
• Performed local and global conformational change simulations.

Main Results:

• The hydrophobic penalty function could distinguish correctly folded structures from incorrect ones.
• Local conformational changes allowed generation of native-like structures efficiently.
• Global simulations yielded non-native conformations with penalty values similar to the native structure.
• These non-native structures, though compact, lacked essential secondary structures.

Conclusions:

• The hydrophobic penalty function alone is insufficient to uniquely define the native protein structure.
• Substructure information is necessary to guide the penalty function towards the correct fold.
• Further refinement of prediction algorithms incorporating structural motifs is required.