Ligand expansion in ligand-based virtual screening using relevance feedback.

Area of Science:

• Computational chemistry
• Cheminformatics
• Drug discovery

Background:

• Information retrieval systems often employ query expansion to enhance search performance.
• Relevance feedback is a key technique for query modification in information retrieval.

Purpose of the Study:

• To integrate query expansion into ligand-based virtual screening (LBVS) using relevance feedback.
• To improve the efficiency and effectiveness of identifying potential drug candidates.

Main Methods:

• A Bayesian inference network was used with a single ligand molecule to generate initial results.
• High-ranking molecules of unknown activity were filtered to create a set of ligands.
• This ligand set was then used to generate a new, expanded query for LBVS.

Main Results:

• Ligand expansion significantly improved LBVS performance, particularly for structurally heterogeneous active molecules.
• The method offers a simple approach to enhance virtual screening outcomes.
• Effectiveness was slightly reduced when searching for structurally homogeneous active compounds.

Conclusions:

• Query expansion with relevance feedback is a valuable technique for enhancing ligand-based virtual screening.
• This approach effectively improves the identification of diverse active molecules.
• The method provides a straightforward yet powerful tool for drug discovery efforts.