Multiple-ligand-based virtual screening: methods and applications of the MTree approach

Area of Science:

• Computational chemistry and cheminformatics
• Drug discovery and medicinal chemistry

Background:

• Ligand-based virtual screening is crucial for identifying novel drug candidates.
• Existing methods using single molecular descriptors can limit the scope of identified leads.

Purpose of the Study:

• To introduce a novel multiple feature tree (MTree) model for enhanced ligand-based virtual screening.
• To improve the identification of new lead structures and alternative molecular scaffolds.
• To identify key molecular features influencing target affinity.

Main Methods:

• Molecules are represented using the established feature tree descriptor derived from topological molecular graphs.
• A novel pairwise alignment algorithm ensures chemically consistent topological molecular alignment.
• Multiple feature tree models (MTree) are constructed by combining query molecules.

Main Results:

• Retrospective virtual screening using MTree models for angiotensin-converting enzyme and alpha1a receptor achieved high enrichment factors (up to 71 in the top 1%).
• MTree models demonstrated superior performance over single feature tree searches in hit rates and quality.
• The methodology successfully identified novel molecular scaffolds not present in the initial query set.

Conclusions:

• The MTree approach offers a significant advancement in ligand-based virtual screening.
• This method effectively identifies promising lead structures and diverse chemical scaffolds for optimization.
• MTree facilitates the discovery of critical molecular features associated with target binding affinity.