High-Throughput Virtual Screening

This tutorial demonstrates high-throughput virtual screening using AutoDock Vina to screen compound libraries against protein targets.

Tutorial Overview

Learning Objectives

By completing this tutorial, you will:

• Master high-throughput molecular docking workflows
• Understand drug repurposing strategies and applications
• Optimize screening parameters for speed vs accuracy
• Analyze large-scale docking results with interactive dashboards

Tutorial Example

Prerequisites

Tutorial Dataset

We'll use a comprehensive screening setup:

Target Information

GLP-1 Receptor

• Type: G-protein coupled receptor (GPCR)
• Structure: PDB 7S15 with co-crystallized ligand
• Applications: Diabetes and obesity treatments
• Binding Site: Deep orthosteric pocket

Stage 1: Target Preparation and Analysis

Setting Up the Screening Target

Stage 2: High-Throughput Screening Setup

Configuring AutoDock Vina for Screening

Add AutoDock Vina node configured for high-throughput:

Parameter	Screening Value	Explanation
Exhaustiveness	8	Actual parameter from analysis
Number of Poses	5	Reduce output size, focus on best pose
Energy Range	5 kcal/mol	Include moderately active compounds
Search Box	25Å × 25Å × 25Å	Actual coordinates from analysis

Screening Strategy Overview

Screening Process

• Screen all 4,278 FDA-approved compounds
• AutoDock Vina with exhaustiveness=8
• Generate binding affinities and poses
• Export results for analysis

Screening Execution

AutoDock Vina processes all compounds automatically and generates results for analysis.

Stage 3: Docking Report Generation

Screening Results Dashboard

The system generates an interactive HTML dashboard with comprehensive analysis:

Summary Metrics

• Total compounds: 4,278 FDA-approved drugs
• Success rate: 98.9% (successful dockings)
• Best affinity: -23.12 kcal/mol
• Binding site: (69.69, 69.69, 60.66)

Hit Classification

• Excellent hits (< -10 kcal/mol): 2,925 compounds (69.7%)
• Good hits (-10 to -8 kcal/mol): 683 compounds (16.3%)
• Moderate hits (-8 to -6 kcal/mol): 371 compounds (8.8%)
• Weak hits (≥ -6 kcal/mol): 219 compounds (5.2%)

Statistical Analysis

• Mean affinity: -11.55 ± 3.45 kcal/mol
• Median affinity: -11.63 kcal/mol
• Enrichment factor (top 1%): 6.5x fold enrichment

Interactive Dashboard Features

Visualizations

• Affinity distribution plots: Histogram of binding scores
• Top hits ranking: Best compounds with Z-scores
• Enrichment analysis: Hit concentration in top compounds
• Statistical summaries: Mean, median, quartiles

Data Tables

• Detailed results: All compounds ranked by affinity
• Quality assessment: Z-scores and percentile rankings
• Compound information: Names, binding modes, statistics
• Export options: Download data for further analysis

Stage 4: Mol* Visualization

3D Structure Analysis

For detailed analysis of top-ranked compounds:

• Binding pose visualization: View compound orientations in active site
• Interaction analysis: Key protein-ligand contacts
• Comparison mode: Side-by-side evaluation of multiple hits

Best Practices

• Use well-validated target structures
• Ensure binding site is properly defined
• Focus on compounds with strong binding affinity
• Validate binding poses visually
• Plan experimental follow-up studies

Conclusion

This tutorial demonstrates high-throughput virtual screening using AutoDock Vina to screen 4,278 FDA-approved drugs against the GLP-1 receptor, achieving excellent hit rates with 2,925 strong binders.

Key Results

• 98.9% success rate with 4,278 compounds screened
• Best affinity: -23.12 kcal/mol
• 69.7% excellent hits (< -10 kcal/mol)
• 6.5x enrichment factor in top compounds

Next Steps

• Explore other applications: DiffDock for novel scaffolds
• Learn protein preparation: Structure Prediction
• Validate hits: GROMACS for binding studies

Resources and References

Key Publications

1. Trott, O. & Olson, A.J. (2010). AutoDock Vina: improving the speed and accuracy of docking with a new scoring function. J Comput Chem 31, 455-461.
2. Sehnal, D. et al. (2021). Mol* Viewer: modern web app for 3D visualization and analysis of large biomolecular structures. Nucleic Acids Research, 49(W1), W431-W437.
3. GitHub repositories:
   • AutoDock Vina: https://github.com/ccsb-scripps/AutoDock-Vina
   • Mol*: https://github.com/molstar/molstar

Databases and Resources

• DrugCentral: https://drugcentral.org/ (FDA-approved drugs)
• PDB: https://www.rcsb.org/ (protein structures)

This tutorial demonstrates high-throughput virtual screening workflows using AutoDock Vina for computational hit identification and analysis.