Adaptive GPU-accelerated force calculation for interactive rigid molecular docking using haptics.

Area of Science:

• Computational chemistry
• Molecular modeling
• Biophysics

Background:

• Molecular docking simulates intermolecular binding interactions in silico.
• Haptics-assisted docking allows user interaction via touch, but requires high computational speeds for feedback.
• Current methods face computational limitations for real-time, high-fidelity haptic feedback.

Purpose of the Study:

• To develop an adaptive, parallelized force calculation method for haptics-assisted molecular docking.
• To achieve high-frequency force updates (500Hz-1kHz) for stable haptic feedback.
• To enhance the applicability and performance of interactive molecular simulations.

Main Methods:

• An adaptive force calculation approach utilizing Graphics Processing Units (GPUs) for parallel processing.
• Selection of either a regular grid or an octree based on GPU memory for interaction calculations.
• Calculation of total force from interatomic interactions within a defined cutoff distance.

Main Results:

• Achieved force updates in under 2ms for large molecular structures (hundreds of thousands of atoms).
• Demonstrated performance improvements up to 90 times faster than current CPU-based methods.
• Overcame limitations of previous methods like pre-computed force grids.

Conclusions:

• The developed GPU-accelerated approach enables interactive haptics-assisted molecular docking with high fidelity.
• The method significantly enhances computational efficiency for molecular simulations.
• Potential applications include modeling receptor flexibility at haptic refresh rates.