Efficiently sampling conformations and pathways using the concurrent adaptive sampling (CAS) algorithm.

Area of Science:

• Computational chemistry
• Biophysics
• Molecular dynamics

Background:

• Molecular dynamics (MD) simulations are crucial for studying biomolecular properties but face time scale limitations.
• Enhanced sampling methods like the weighted ensemble (WE) method exist but require predefined phase space partitioning.
• High-dimensional systems or unknown reaction coordinates pose challenges for traditional WE methods.

Purpose of the Study:

• To introduce the Concurrent Adaptive Sampling (CAS) algorithm, a novel WE-based method.
• To address the limitations of existing methods in handling high-dimensional collective spaces and unknown reaction coordinates.
• To enhance the efficiency of sampling thermodynamic and kinetic properties in complex molecular systems.

Main Methods:

• Developed the Concurrent Adaptive Sampling (CAS) algorithm, a WE-based enhanced sampling technique.
• CAS utilizes a large number of collective variables and adaptive macrostates, overcoming the limitations of single or few collective variables.
• Incorporated a clustering technique based on the committor function to accelerate the sampling of slow molecular processes.

Main Results:

• CAS demonstrates enhanced sampling capabilities in high-dimensional spaces.
• The method effectively handles systems where optimal reaction coordinates are not initially known.
• Successful application of CAS to two-dimensional models, penta-alanine, and a triazine trimer.

Conclusions:

• The CAS algorithm offers a robust solution for overcoming time scale barriers in molecular simulations.
• CAS provides efficient sampling for complex biomolecular systems, particularly those with high-dimensional conformational landscapes.
• This method advances the study of molecular dynamics and kinetics by enabling exploration of previously inaccessible time scales and conformational spaces.