Biomolecular dynamics at long timesteps: bridging the timescale gap between simulation and experimentation

Area of Science:

• Computational Chemistry
• Molecular Dynamics
• Biophysics

Background:

• Simulating Newtonian physics for macromolecules requires small timesteps for numerical stability, limiting simulation length.
• Existing faster methods struggle with strong vibrational coupling in biomolecules.
• Bridging the gap to biologically relevant timescales (milliseconds) remains a challenge.

Purpose of the Study:

• To review algorithms for accelerating molecular dynamics simulations.
• To examine methods that alleviate the timestep constraint in simulating macromolecular motion.
• To introduce and evaluate a novel dual timestep method (LN).

Main Methods:

• Review of existing timestep alleviation techniques (constrained, reduced-variable, implicit, symplectic, multiple-timestep, normal-mode-based).
• Comparison of integrator performance on a model dipeptide.
• Implementation and assessment of the dual timestep LN method with approximate linearization.

Main Results:

• The LN method achieves a 4-5x computational speedup compared to standard 0.5 fs trajectories.
• LN results show good agreement with high-resolution simulations.
• The method effectively addresses the timestep limitation for longer molecular dynamics simulations.

Conclusions:

• Algorithmic advancements, including the LN method, are crucial for simulating biologically relevant timescales.
• The LN method offers a computationally competitive approach to molecular dynamics.
• Further development and integration with experimental data are needed for theoretical modeling to match experimental capabilities.