AQDnet: Deep Neural Network for Protein-Ligand Docking Simulation.

Area of Science:

• Computational chemistry
• Structural biology
• Drug discovery

Background:

• Accurate prediction of protein-ligand binding affinity is crucial for drug discovery.
• Existing methods face challenges in capturing the complex interactions within protein-ligand complexes.
• Quantum mechanics calculations offer high accuracy but are computationally expensive for large datasets.

Purpose of the Study:

• To develop an innovative system, AI QM Docking Net (AQDnet), for predicting protein-ligand binding affinity.
• To enhance the training dataset by generating diverse ligand configurations and calculating binding energies via quantum computation.
• To incorporate atom-centered symmetry functions (ACSF) for improved prediction of protein-ligand interactions.

Main Methods:

• Generation of thousands of diverse ligand configurations for each protein-ligand complex.
• Determination of binding energy for each configuration using quantum computation.
• Development of a neural network trained on the protein-ligand quantum energy landscape (P-L QEL) using ACSF.

Main Results:

• Achieved a 92.6% top 1 success rate in the CASF-2016 docking power benchmark.
• AQDnet outperformed all previously assessed models in the CASF-2016 evaluation.
• Demonstrated the effectiveness of ACSF in predicting protein-ligand interactions.

Conclusions:

• AQDnet represents a significant advancement in predicting protein-ligand binding affinity.
• The system's novel approach to dataset expansion and ACSF integration leads to exceptional docking performance.
• This model holds great promise for accelerating drug discovery and development processes.