Advanced techniques for constrained internal coordinate molecular dynamics.

Area of Science:

• Computational chemistry
• Biophysics
• Molecular dynamics

Background:

• Internal coordinate molecular dynamics (ICMD) offers a more natural protein representation than Cartesian coordinates.
• Constrained ICMD (CICMD) models arise from freezing high-frequency bonds/angles, but require theoretical development for wider usability.

Purpose of the Study:

• To develop a robust framework for constrained internal coordinate molecular dynamics (CICMD) simulations.
• To enhance the study of protein dynamics, structure prediction, and refinement.

Main Methods:

• Designed a new framework for initializing velocities, computing center of mass velocity, and using advanced integrators (Runge-Kutta, Lobatto, CVODE) in CICMD.
• Implemented the Generalized Newton-Euler Inverse Mass Operator (GNEIMO) method for CICMD simulations.
• Developed an adaptive coarse-graining tool using the GNEIMO Python interface for on-the-fly freezing/thawing of degrees of freedom.

Main Results:

• Successfully folded four proteins to their native topologies using the adaptive coarse-graining tool.
• Demonstrated the ability to constrain any group of atoms, enabling a hierarchy of coarse-grained models.
• Addressed theoretical aspects including velocity initialization, efficient center of mass velocity computation, advanced integrators, and cancellation of the 'flying ice cube effect'.

Conclusions:

• The GNEIMO method provides a powerful and flexible platform for CICMD simulations.
• Advancements in GNEIMO facilitate accurate protein folding, structure prediction, refinement, and domain motion studies.
• The developed adaptive coarse-graining tool significantly enhances the utility of molecular dynamics simulations.