Binding pose and affinity prediction in the 2016 D3R Grand Challenge 2 using the Wilma-SIE method

Area of Science:

• Molecular pharmacology
• Computational chemistry
• Structural biology

Background:

• The Farnesoid X receptor (FXR) undergoes significant conformational changes upon ligand binding, posing challenges for molecular docking and pose prediction algorithms.
• Accurate prediction of ligand binding poses and affinities is crucial for drug discovery and development.

Purpose of the Study:

• To evaluate the performance of the Wilma-SIE rigid-protein docking method for predicting poses of 36 Farnesoid X receptor (FXR) ligands.
• To assess the impact of using an ensemble of FXR protein structures versus a single structure on pose prediction accuracy.
• To investigate the accuracy of affinity predictions for FXR ligands.

Main Methods:

• Utilized the Wilma-SIE rigid-protein docking method within the D3R Grand Challenge 2 framework.
• Employed an ensemble of publicly available Farnesoid X receptor (FXR) structures from the Protein Data Bank (PDB) to overcome rigid-protein limitations.
• Tested the method on 36 blinded FXR ligands with known crystal structures.

Main Results:

• Achieved average and median RMSD values of 2.3 and 1.4 Å, respectively, for pose predictions using the FXR ensemble.
• Demonstrated that using a single receptor structure significantly reduced prediction success rates compared to an ensemble approach.
• Observed improved prediction accuracy for ligand classes with available co-crystal structures and identified consensus binding modes as indicators of reasonable poses in their absence.
• Reported generally poor correlation between predicted and experimental affinities (Kendall's tau ~0.3), with minimal improvement even when using all 36 crystal structures.

Conclusions:

• An ensemble of protein structures is essential for accurate pose prediction with rigid-docking methods like Wilma-SIE when dealing with flexible receptors such as FXR.
• Affinity prediction for FXR ligands remains a significant challenge, potentially due to the need to accurately model internal energy strain from receptor conformational flexibility.
• Future efforts should focus on incorporating receptor conformational dynamics into scoring functions to improve affinity prediction accuracy.