Thermodynamic resolution: how do errors in modeled protein structures affect binding affinity predictions?

Area of Science:

• Computational chemistry and structural biology.
• Drug discovery and development.

Background:

• Protein structure modeling is crucial for understanding molecular interactions.
• Errors in protein models can arise from various computational methods.
• Accurate binding affinity prediction is essential for identifying drug candidates.

Purpose of the Study:

• To quantify the effect of structural distortion in protein models on calculated binding affinities.
• To establish a relationship between protein structure errors and binding affinity prediction errors.
• To assess the reliability of modeled protein structures for drug development.

Main Methods:

• Evaluation of binding affinities for 300 distorted protein structures across five protein-ligand complexes.
• Systematic analysis of binding affinity deviations based on root-mean-square deviation (RMSD) of protein models.
• Comparison of binding affinities calculated from distorted models versus crystal structures.

Main Results:

• Protein models with <2 Å RMSD showed a 14.78% deviation in binding affinity.
• Increasing RMSD values (2-3 Å, 3-4 Å, >4 Å) led to progressively higher affinity errors (20.8%, 22.79%, 29.43%).
• A clear correlation exists between the degree of structural distortion and the magnitude of binding affinity error.

Conclusions:

• Structural distortions in protein models significantly compromise the accuracy of binding affinity calculations.
• The findings provide a quantitative basis for evaluating the quality of modeled protein structures in ligand-binding studies.
• This research aids in optimizing the use of computational models for efficient drug discovery.