Prediction of protein side-chain conformation by packing optimization

Area of Science:

• Computational biology
• Structural bioinformatics
• Protein structure prediction

Background:

• Predicting protein side-chain conformation is crucial for understanding protein function and interactions.
• Current methods struggle with the combinatorial complexity of simultaneously predicting multiple residue conformations.
• Accurate prediction of core residues is particularly challenging yet vital for protein stability.

Purpose of the Study:

• To develop a rapid, automated method for predicting protein side-chain conformation.
• To optimize side-chain packing interactions using simulated annealing.
• To address the combinatorial challenge in predicting conformations for numerous residues.

Main Methods:

• Application of simulated annealing algorithm for optimizing van der Waals interactions.
• Directly tackling the combinatorial problem of simultaneous residue conformation prediction.
• Testing the method on nine proteins of varying sizes (46–323 residues).

Main Results:

• Achieved an overall root-mean-square (r.m.s.) deviation of 1.77 Å for side-chain atoms compared to native structures.
• Demonstrated high accuracy for core residues with an overall r.m.s. of 1.25 Å.
• Correctly predicted 80–90% of large hydrophobic core side-chains, outperforming current methods (30–40%).

Conclusions:

• Van der Waals interactions are the primary drivers of side-chain arrangement in protein cores.
• The developed method accurately predicts native-like packing, especially for core residues.
• Future improvements could incorporate surface residue constraints and hydrogen bonding interactions.