GPU-Accelerated Implementation of Constant-pH Molecular Dynamics in NAMD.

Area of Science:

• Computational Biology
• Biophysics
• Molecular Dynamics

Background:

• Simulating biomolecular systems at a defined pH is crucial for understanding their function.
• Conventional molecular dynamics methods face challenges in handling protonation state changes.
• Existing constant-pH simulations can be computationally intensive.

Purpose of the Study:

• To introduce a GPU-accelerated implementation of the hybrid nonequilibrium molecular dynamics-Monte Carlo constant-pH simulation method.
• To enhance the efficiency and applicability of all-atom simulations for biomolecules at a specific pH.
• To enable more comprehensive investigations of dynamic protonation profiles.

Main Methods:

• Implemented a hybrid nonequilibrium molecular dynamics-Monte Carlo (MD-MC) constant-pH simulation approach on Graphics Processing Units (GPUs).
• Integrated the GPU-based method into the NAMD simulation package.
• Conducted benchmark tests comparing GPU and CPU performance.

Main Results:

• The GPU-based implementation achieved significant speedup compared to the CPU-based version.
• The accuracy of the simulations was maintained at the same level as the CPU counterpart.
• The performance improvement broadens the scope of pH-controlled biomolecular simulations.

Conclusions:

• GPU-acceleration of constant-pH MD-MC simulations offers a powerful and efficient approach.
• This advancement facilitates the study of dynamic protonation states in diverse biomolecular systems.
• The method overcomes limitations of exhaustive protonation-state enumeration in traditional molecular dynamics.