Bridging Structure- and Ligand-Based Virtual Screening through Fragmented Interaction Fingerprint.

Area of Science:

• Computational chemistry
• Medicinal chemistry
• Drug discovery

Background:

• Ligand-based virtual screening (LBVS) and structure-based virtual screening (SBVS) are key in drug discovery.
• Hybrid VS approaches combine ligand and structure information for better compound prioritization.
• Interaction fingerprints (IFPs) are used in hybrid VS, but novel methods are needed.

Purpose of the Study:

• To introduce a new IFP, fragmented interaction fingerprint (FIFI), for hybrid VS.
• To evaluate FIFI's performance against existing IFPs and other VS strategies.
• To assess the efficacy of hybrid VS approaches in prioritizing active drug compounds.

Main Methods:

• FIFI construction based on ligand substructures and protein binding site interactions.
• Retrospective evaluation of FIFI and other VS methods on six biological targets.
• Comparison of FIFI-ML with LBVS, SBVS, sequential VS, parallel VS, and other hybrid VS approaches.

Main Results:

• FIFI demonstrated higher prediction accuracy than previous IFPs across most targets.
• FIFI combined with machine learning (ML) showed stable and high prediction accuracy.
• The extended connectivity fingerprint with ML outperformed other methods for the kappa opioid receptor.

Conclusions:

• FIFI represents a promising advancement for hybrid virtual screening in drug discovery.
• Hybrid VS approaches, particularly FIFI-ML, offer robust performance in identifying active compounds.
• Specific targets may benefit from tailored VS methods, like extended connectivity fingerprints for the kappa opioid receptor.