Protein-ligand docking against non-native protein conformers

Area of Science:

• Computational chemistry
• Structural biology
• Drug discovery

Background:

• Protein-ligand docking protocols are typically validated against native protein conformations.
• Real-world applications often involve docking ligands into non-native protein structures (e.g., apo or different complexes).
• This discrepancy between validation and application settings can impact docking accuracy.

Purpose of the Study:

• To develop and utilize an extensive test set for assessing protein-ligand docking performance against non-native protein conformations.
• To quantify the performance drop-off in docking when using non-native protein structures compared to native ones.
• To investigate strategies for improving docking performance in non-native scenarios.

Main Methods:

• Construction of the Astex Non-native Set, comprising 1112 non-native structures for 65 drug targets, based on the Astex Diverse Set.
• Utilizing the GOLD protein-ligand docking program to evaluate docking performance on both native and non-native structure sets.
• Analysis of docking performance (top-ranked solution accuracy) and sampling performance (any correct solution).

Main Results:

• Docking performance dropped from approximately 80% for native docking to 61% for non-native docking.
• Sampling performance decreased from 91% for native to 72% for non-native docking.
• Larger protein conformational changes significantly degraded performance, while docking against similar ligands or multiple non-native structures improved results.

Conclusions:

• Current protein-ligand docking validation methods may overestimate performance due to the use of native protein conformations.
• Docking against non-native protein conformations presents a significant challenge, necessitating the use of more realistic test sets.
• Strategies like ensemble docking or docking against similar ligand complexes can mitigate performance loss in non-native scenarios.