Protein side-chain packing problem: is there still room for improvement?

Area of Science:

• Computational Biology
• Biophysics
• Structural Bioinformatics

Background:

• The protein side-chain packing problem (PSCPP) is a critical component in protein structure prediction and design.
• Numerous computational methods have been developed over the last two decades to address PSCPP.
• Existing methods typically comprise a rotamer library, a scoring function, and a search strategy, achieving an average accuracy of approximately 87%.

Purpose of the Study:

• To determine the maximum achievable accuracy for PSCPP using a basic rotamer library, independent of specific energy functions or search algorithms.
• To assess the gap between current method performance and theoretical maximum accuracy, particularly for buried versus exposed residues.
• To identify the primary limitations hindering improved accuracy in PSCPP.

Main Methods:

• Calculated the theoretical maximum accuracy for PSCPP using a simple rotamer library.
• Compared this maximum accuracy with the performance of five state-of-the-art methods on 2883 Protein Data Bank structures.
• Evaluated the accuracy of two commonly used scoring functions in guiding search strategies for PSCPP.

Main Results:

• For buried residues, current methods are nearing the maximum achievable accuracy.
• A significant gap remains for exposed residues, indicating potential for improvement.
• Two evaluated scoring functions demonstrated inaccuracies, failing to guide search methods effectively.

Conclusions:

• While buried residue packing is near optimal, exposed residue packing in PSCPP offers room for advancement.
• Scoring function deficiencies are a major bottleneck preventing higher accuracy in PSCPP.
• Future efforts to enhance PSCPP should prioritize the development of more accurate scoring functions.