Related Experiment Video
Updated: Jun 5, 2026
Biosensor-based High Throughput Biopanning and Bioinformatics Analysis Strategy for the Global Validation of Drug-protein Interactions
Published on: December 1, 2020
Mining flexible-receptor docking experiments to select promising protein receptor snapshots
Karina S Machado1, Ana T Winck, Duncan D A Ruiz
1LABIO - Laboratório de Bioinformática, Modelagem e Simulação de Biossistemas, PPGCC, Faculdade de Informática, PUCRS, Av. Ipiranga, 6681 - Prédio 32, sala 602, 90619-900, Porto Alegre, RS, Brazil.
This study introduces a method to efficiently select promising protein receptor conformations for drug design by analyzing molecular docking data. This approach accelerates rational drug design by identifying key receptor snapshots, improving efficiency in identifying potential drug candidates.
Area of Science:
- Computational Chemistry
- Structural Biology
- Drug Discovery
Background:
- Molecular docking is crucial for Rational Drug Design (RDD), but receptor flexibility is often ignored or computationally expensive to model.
- Traditional docking treats receptors as rigid, limiting accuracy in predicting ligand-receptor interactions.
- Exploring receptor flexibility using multiple conformations is time-consuming, necessitating efficient selection methods.
Purpose of the Study:
- To develop a method for selecting the most promising protein receptor conformations from a large set.
- To accelerate molecular docking simulations and enhance the efficiency of Rational Drug Design (RDD) efforts.
- To identify key receptor snapshots that accurately represent ligand-receptor binding.
Main Methods:
- Docking of four ligands (NADH, TCL, PIF, ETH) to 3,100 conformations of the M. tuberculosis InhA receptor.
- Preprocessing docking results to calculate shortest interatomic distances between ligands and receptor residues.
- Utilizing the M5P model tree algorithm for data mining and identifying predictive attributes.
- Post-processing generated model trees to select representative linear models based on average free energy of binding (FEB).
Main Results:
- High correlation (>95%) between predicted and actual FEB for NADH, TCL, and PIF ligands.
- The M5P model tree algorithm generated concise and interpretable models.
- A criterion was established to select linear models, identifying 1,521 to 2,085 promising receptor conformations (snapshots) per ligand.
- Successful identification of representative receptor conformations that accelerate docking experiments.
Conclusions:
- Post-processing model trees provides a robust criterion for selecting promising receptor conformations.
- This method significantly accelerates molecular docking and RDD by focusing on relevant receptor snapshots.
- Future work includes developing strategies for receptor 3D conformation preprocessing to predict FEB values and identify novel drug candidates.
