Docking-undocking combination applied to the D3R Grand Challenge 2015

Area of Science:

• Computational chemistry
• Drug discovery
• Structural biology

Background:

• Novel drug discovery methods require rigorous validation.
• The Drug Discovery Data Resource (D3R) Grand Challenge 2015 provided a platform for assessing computational methods.
• Two protein targets, Hsp90 and MAP4K4, were included in the challenge.

Purpose of the Study:

• To externally assess and validate Computer-Aided Drug Discovery (CADD) methods.
• To evaluate the efficacy of pharmacophore-guided docking combined with dynamic undocking.
• To investigate the impact of prior knowledge on prediction accuracy.

Main Methods:

• Employed pharmacophore-guided docking followed by dynamic undocking.
• Utilized a strategy combining computational methods with critical assessment of pre-existing information.
• Tested methods on binding mode and ligand ranking prediction tests for two protein targets.

Main Results:

• Achieved top-tier results for binding mode and ligand ranking on the Hsp90 target.
• Observed less positive results for the MAP4K4 target, correlating with limited prior knowledge of its conformational states.
• Demonstrated the effectiveness of docking when augmented by dynamic undocking and empirical data.

Conclusions:

• Docking, when supplemented with dynamic undocking and empirical information, is effective for drug discovery.
• Prior knowledge of protein conformational states significantly influences prediction accuracy.
• Protein flexibility remains a key challenge in computational drug discovery, leading to potential failures.