Reducing the number of mean-square deviation calculations with floating close structure in metadynamics

Area of Science:

• Computational Chemistry
• Molecular Dynamics
• Biophysics

Background:

• Metadynamics is a key enhanced sampling method in molecular simulations.
• Its effectiveness relies on well-chosen collective variables (CVs).
• Landmark-based CVs improve sampling in complex systems but are computationally intensive due to frequent distance calculations (e.g., root-mean-square deviation).

Purpose of the Study:

• To develop and validate an approximation for root-mean-square deviation (RMSD) calculations.
• To reduce the computational cost of landmark-based collective variables in metadynamics.
• To improve the efficiency of enhanced sampling simulations for complex molecular systems.

Main Methods:

• Introduction of a novel approximation for RMSD calculation.
• Integration of the approximated RMSD into metadynamics simulations.
• Validation of the approximation's accuracy and performance gain on two molecular systems.

Main Results:

• The proposed approximation significantly reduces the number of computationally expensive operations.
• Metadynamics simulations using the approximated RMSD show comparable accuracy to standard methods.
• Theoretical performance gains were evaluated and confirmed through simulations.

Conclusions:

• The developed RMSD approximation offers a computationally efficient alternative for landmark-based CVs in metadynamics.
• This method can accelerate enhanced sampling simulations of complex molecular processes.
• The approach holds potential for broader application in molecular modeling and simulation studies.