AutoDock-SS: AutoDock for Multiconformational Ligand-Based Virtual Screening.

Area of Science:

• Computational chemistry
• Drug discovery
• cheminformatics

Background:

• Ligand-based virtual screening (LBVS) is crucial for identifying drug leads when protein structures are unavailable.
• Existing LBVS methods often fail to adequately account for ligand conformational flexibility, limiting their effectiveness.

Purpose of the Study:

• To introduce AutoDock-SS (Similarity Searching), a novel 3D LBVS workflow that integrates ligand-based grid maps and AutoDock-GPU.
• To address the limitations of current LBVS methods by incorporating dynamic conformational searching.

Main Methods:

• Developed AutoDock-SS, a workflow adapting protein-ligand docking for LBVS.
• Integrated novel ligand-based grid maps and AutoDock-GPU.
• Implemented a built-in conformational search for dynamic optimization of ligand conformations.
• Supported single and multiple ligand query modes.

Main Results:

• AutoDock-SS achieved a mean AUROC of 0.775 and EF1% of 25.72 in single-reference mode on the DUD-E dataset.
• In multireference mode on DUD-E+, AutoDock-SS demonstrated superior accuracy with a mean AUROC of 0.843 and EF1% of 34.59.
• Outperformed alternative 3D LBVS methods in both tested modes.

Conclusions:

• AutoDock-SS significantly enhances LBVS by treating ligands as conformationally flexible.
• The method accurately considers ligand shape, pharmacophore, and electrostatic potential.
• AutoDock-SS expands the capabilities and potential applications of LBVS in drug discovery.