Building multiscale Markov state models by systematic mapping of temporal communities.

Area of Science:

• Computational Biology
• Biophysics
• Molecular Dynamics

Background:

• Biomolecules function through dynamic transitions between metastable states.
• Markov State Models (MSMs) analyze these transitions at a single temporal scale.
• Biological processes involve dynamics across a wide range of timescales.

Purpose of the Study:

• To introduce a method for analyzing biomolecular dynamics across multiple temporal scales simultaneously.
• To develop an algorithm for generating these multiscale models.
• To demonstrate the capability of the method in mapping complex systems.

Main Methods:

• Development of multiscale Markov State Models (mMSMs) using a hierarchy of MSMs.
• Implementation of mMSM-explore, an unsupervised algorithm for adaptive sampling.
• On-the-fly identification of temporally metastable states.
• Benchmarking on toy systems, alanine dipeptide, and a miniprotein.

Main Results:

• Efficient mapping of energy landscapes and multiscale hierarchies.
• Accurate representation of transition states and kinetics.
• De novo identification of slow, intermediate, and fast reaction coordinates.
• Demonstration of collective contributions of multiscale dynamics to function.

Conclusions:

• mMSMs provide a comprehensive framework for understanding biomolecular dynamics across timescales.
• The mMSM-explore algorithm enables efficient generation and analysis of these models.
• This approach enhances our understanding of the functional mechanisms of biomolecular machines.