PLHINT: A knowledge-driven computational approach based on the intermolecular H bond interactions at the...

Area of Science:

• Computational chemistry
• Structural biology
• Drug discovery

Background:

• Docking scoring functions often fail to reproduce crystal intermolecular contacts.
• Protein-ligand H-bond interactions are crucial for accurate binding pose prediction.
• Existing methods need enhancement to incorporate detailed H-bond information into docking solutions.

Purpose of the Study:

• To develop a computational approach, PLHINT (Protein-ligand H bond interaction pattern), to improve virtual screening.
• To integrate knowledge of protein-ligand H-bond contacts into docking scoring schemes.
• To enhance the recovery rates and enrichment competency of virtual screening.

Main Methods:

• PLHINT extracts intermolecular H-bond interactions from native-like docked poses.
• Identified H-bond patterns are transformed into a scoring scheme.
• The PLHINT scoring scheme is applied to virtual screening results of database molecules.
• The approach was tested on ten diverse DUD-E benchmark datasets.

Main Results:

• PLHINT demonstrated improved overall performance and ligand enrichment compared to standard docking programs (AutoDock Vina, FlexAID, PLANTS).
• The method successfully improved poor and random virtual screening results.
• High recovery rates were achieved by scoring frequently observed H-bond patterns.

Conclusions:

• PLHINT effectively incorporates protein-ligand H-bond information to enhance virtual screening.
• The approach offers a valuable tool for improving the accuracy and efficiency of drug discovery pipelines.
• PLHINT addresses limitations in current docking scoring functions by leveraging specific interaction patterns.