Correlation between Virtual Screening Performance and Binding Site Descriptors of Protein Targets

Area of Science:

• Computational chemistry
• Drug discovery
• Bioinformatics

Background:

• Molecular docking is crucial for identifying potential drug candidates.
• AutoDock Vina is a widely used docking engine.
• Rescoring can potentially enhance the accuracy of docking predictions.

Purpose of the Study:

• To evaluate the impact of rescoring on virtual screening results obtained from AutoDock Vina.
• To explore the effectiveness of different scoring functions, including consensus approaches.
• To develop a predictive model for rescoring improvements.

Main Methods:

• Utilized AutoDock Vina as the primary docking engine.
• Employed three additional scoring functions and their combinations for rescoring.
• Investigated diverse biological targets for virtual screening.
• Developed a predictive model based on two binding site descriptors.

Main Results:

• Rescoring with DrugScore demonstrated the most significant changes in screening power across various targets.
• A model was built to predict potential improvements from DrugScore rescoring.
• The screening power of rescoring methods and AutoDock Vina correlated with Maximum Theoretical Shape Complementarity (MTSC) and Maximum Distance from Center of Mass and all Alpha spheres (MDCMA).

Conclusions:

• Rescoring, particularly with DrugScore, can substantially influence virtual screening outcomes.
• Binding site descriptors show promise in predicting the effectiveness of rescoring strategies.
• Further research with comprehensive descriptors may establish robust relationships between binding site characteristics and docking program performance.