New fragment weighting scheme for the Bayesian inference network in ligand-based virtual screening

Area of Science:

• Computational Chemistry
• Cheminformatics
• Drug Discovery

Background:

• Conventional molecular similarity methods often assign equal importance to all molecular fragments, regardless of their relevance to biological activity.
• Bayesian inference networks (BINs) offer a probabilistic approach to model molecular relationships, addressing limitations of traditional methods.

Purpose of the Study:

• To develop and evaluate a new fragment weighting scheme for Bayesian inference networks (BINs) to enhance molecular similarity assessment.
• To improve the accuracy and effectiveness of virtual screening in drug discovery.

Main Methods:

• Investigated five different weighting functions for BINs, incorporating a novel fragment weighting scheme.
• Modified existing functions to integrate the new weighting scheme.
• Conducted simulated virtual screening experiments using the MDL Drug Data Report and maximum unbiased validation datasets.

Main Results:

• The new fragment weighting scheme significantly improved the effectiveness of virtual screening.
• The enhanced BIN approach demonstrated superior performance compared to current weighting schemes in identifying active compounds.

Conclusions:

• The proposed fragment weighting scheme offers a more effective approach for molecular similarity calculations within BINs.
• This advancement has the potential to enhance the efficiency and success rate of virtual screening in drug discovery pipelines.