Computing committors via Mahalanobis diffusion maps with enhanced sampling data.

Area of Science:

• Chemical Physics
• Computational Biophysics
• Biomolecular Simulations

Background:

• Molecular dynamics (MD) simulations are crucial for studying biomolecular transitions.
• A significant timescale gap hinders MD simulations of large-scale processes.
• Enhanced sampling techniques are needed to bridge this gap.

Purpose of the Study:

• To develop an enhanced algorithm for quantifying large-timescale transition processes in biomolecules.
• To integrate diffusion maps with enhanced sampling methods for improved accuracy.
• To address the limitations of traditional MD simulations in studying protein folding and conformational changes.

Main Methods:

• Developed the target measure Mahalanobis diffusion map (tm-mmap) algorithm.
• Incorporated enhanced sampling techniques like metadynamics into diffusion map analysis.
• Utilized the Backward Kolmogorov Operator (BKO) approximation for committor function calculation.
• Applied transition path theory to determine reactive currents and transition rates.

Main Results:

• The tm-mmap algorithm successfully approximates diffusion tensors and free energy landscapes.
• Validated tm-mmap on a 2D Moro-Cardin system and alanine dipeptide in two collective variables against FEM.
• Demonstrated tm-mmap's feasibility for alanine dipeptide in four collective variables where FEM is intractable.

Conclusions:

• tm-mmap offers a robust method for analyzing complex biomolecular transitions beyond the reach of standard MD.
• The algorithm accurately calculates committor functions, reactive currents, and transition rates.
• tm-mmap provides a powerful tool for chemical physics and computational biophysics research.