Reaction Coordinates for Conformational Transitions Using Linear Discriminant Analysis on Positions.

Area of Science:

• Computational Biology
• Biophysics
• Statistical Mechanics

Background:

• Macromolecular configurations are typically represented by atomic coordinates.
• Traditional enhanced sampling methods often exclude atomic coordinates due to rotational and translational invariance challenges.
• Developing methods to effectively utilize atomic coordinates is crucial for advancing molecular simulations.

Purpose of the Study:

• To demonstrate that Linear Discriminant Analysis (LDA) applied to atomic positions can serve as an effective reaction coordinate.
• To address the limitations of rotational and translational invariance in using atomic coordinates for enhanced sampling.
• To enable accurate free energy estimation for biomolecular transitions.

Main Methods:

• Applying Linear Discriminant Analysis (LDA) to atomic positions of a biomolecule in two distinct states.
• Utilizing a size-and-shape space equivalence class to achieve rotational and translational invariance.
• Performing molecular dynamics (MD) simulations for enhanced sampling and free energy calculations.

Main Results:

• LDA applied to atomic positions effectively generates a reaction coordinate between different biomolecular states.
• The developed method successfully handles rotational and translational invariance.
• The reaction coordinates facilitate accurate characterization of transitions and enable reliable free energy estimation.

Conclusions:

• Linear Discriminant Analysis on atomic positions is a powerful technique for defining reaction coordinates in biomolecular systems.
• This approach enhances the utility of atomic coordinates in enhanced sampling molecular dynamics simulations.
• The method provides a pathway for more precise free energy calculations and understanding molecular mechanisms.