Comparison of ligand- and structure-based virtual screening on the DUD data set

Area of Science:

• Computational chemistry
• Cheminformatics
• Drug discovery

Background:

• Virtual screening is crucial for identifying drug candidates.
• Evaluating the performance of different computational methods is essential for optimizing drug discovery pipelines.
• The Dataset of Useful Decoys (DUD) provides a standardized benchmark for assessing virtual screening tools.

Purpose of the Study:

• To compare the performance of various in-house and commercial computational methods for virtual screening.
• To evaluate the effectiveness of chemical fingerprints, docking tools, and the Flexophore descriptor using the DUD.
• To assess the capability of the Flexophore descriptor for scaffold hopping in drug discovery.

Main Methods:

• Virtual screening was performed using the Dataset of Useful Decoys (DUD), which includes active compounds and structurally similar decoys.
• Four chemical fingerprints, a topological pharmacophore descriptor, the Flexophore descriptor, and the in-house docking tool ActDock were evaluated.
• Enrichment rates and similarity matrix calculations were used to compare the performance of the different methods.

Main Results:

• Chemical fingerprint descriptors demonstrated superior enrichment rates compared to the Flexophore descriptor and the ActDock docking tool.
• After removing chemically similar molecules, the Flexophore descriptor outperformed chemical descriptors and topological pharmacophore descriptors.
• Similarity matrix calculations indicated that the Flexophore descriptor is effective for scaffold hopping, identifying new chemical entities.

Conclusions:

• Chemical fingerprints offer high enrichment rates in virtual screening.
• The Flexophore descriptor shows significant potential for scaffold hopping and discovering novel chemical scaffolds.
• The DUD is a valuable resource for evaluating and comparing virtual screening methodologies.