Improved side-chain prediction accuracy using an ab initio potential energy function and a very large rotamer library

Area of Science:

• Computational Biology
• Structural Bioinformatics
• Protein Science

Background:

• Accurate protein side-chain prediction is crucial for protein design, structure prediction, and functional analysis.
• Current methods often rely on discrete rotamer libraries and scoring functions based on experimental or empirical parameters.
• Existing algorithms face limitations in accuracy and computational efficiency.

Purpose of the Study:

• To introduce the NCN algorithm for predicting protein side-chain placement and conformations.
• To develop a predominantly first-principles potential energy function for side-chain prediction.
• To create the most extensive discrete rotamer library to date.

Main Methods:

• Developed the NCN algorithm utilizing a first-principles approach for its potential energy function.
• Incorporated van der Waals, electrostatics (OPLS parameters), and hydrogen bonding terms.
• Utilized rotameric state frequencies from the Protein Data Bank (PDB) and an extensive library of nearly 50,000 rotamers.

Main Results:

• The NCN algorithm demonstrates superior accuracy compared to existing literature methods for side-chain placement on accurate backbone traces.
• Achieved 92% accuracy for chi(1) and 83% for chi(1 + 2) placement accuracy for buried residues, with an overall RMS deviation of 1 Å.
• When chi(1) is restricted, the algorithm generates structures with an average RMS deviation of 1.0 Å for heavy side-chain atoms and 85.0% chi(1 + 2) accuracy.

Conclusions:

• The NCN algorithm offers a highly accurate method for protein side-chain prediction, exceeding current standards.
• Its physics-based approach and extensive rotamer library contribute to its enhanced predictive power.
• The algorithm holds significant potential for advancing protein design, structure prediction, and functional analysis.