Cloud computing approaches for prediction of ligand binding poses and pathways.

Area of Science:

• Computational chemistry
• Structural biology
• Pharmacology

Background:

• Accurate prediction of ligand binding poses is crucial for drug discovery.
• Analyzing large-scale protein-ligand dynamics data presents computational challenges.

Purpose of the Study:

• To develop an innovative protocol for ab initio prediction of ligand crystallographic binding poses.
• To enable highly effective analysis of large datasets for protein-ligand dynamics.

Main Methods:

• Distributed molecular dynamics simulations on cloud computing architectures.
• A novel model for efficient analysis of simulation data.
• A metric for evaluating model convergence.

Main Results:

• Accurate binding pose predictions for five ligands with varying affinities (7 nM to >200 μM) for FKBP12.
• Expedited results, particularly when experimental structures are difficult to obtain.
• Quantitative kinetic information beyond single low-energy poses.

Conclusions:

• The developed protocol offers a powerful tool for predicting ligand binding poses and analyzing protein-ligand dynamics.
• This approach provides valuable kinetic insights for protein engineering and ligand design.
• Facilitates faster and more efficient drug discovery processes.