Constraint methods for determining pathways and free energy of activated processes

TL;DR

This paper reviews constraint methods for exploring activated chemical and biomolecular processes, focusing on reaction pathways, free energy profiles, and their theoretical and practical aspects, including applications and limitations.

Contribution

It provides a comprehensive introduction and theoretical development of constraint-based methods for calculating reaction pathways and free energy profiles, with practical implementation insights.

Findings

Constraint methods effectively explore difficult pathways.

Free energy profiles can be derived from mean forces.

Applications show good qualitative and quantitative agreement with experiments.

Abstract

Activated processes from chemical reactions up to conformational transitions of large biomolecules are hampered by barriers which are overcome only by the input of some free energy of activation. Hence, the characteristic and rate-determining barrier regions are not sufficiently sampled by usual simulation techniques. Constraints on a reaction coordinate r have turned out to be a suitable means to explore difficult pathways without changing potential function, energy or temperature. For a dense sequence of values of r, the corresponding sequence of simulations provides a pathway for the process. As only one coordinate among thousands is fixed during each simulation, the pathway essentially reflects the system's internal dynamics. From mean forces the free energy profile can be calculated to obtain reaction rates and insight in the reaction mechanism. In the last decade, theoretical tools and computing capacity have been developed to a degree where simulations give impressive qualitative insight in the processes at quantitative agreement with experiments. Here, we give an introduction to reaction pathways and coordinates, and develop the theory of free energy as the potential of mean force. We clarify the connection between mean force and constraint force which is the central quantity evaluated, and discuss the mass metric tensor correction. Well-behaved coordinates without tensor correction are considered. We discuss the theoretical background and practical implementation on the example of the reaction coordinate of targeted molecular dynamics simulation. Finally, we compare applications of constraint methods and other techniques developed for the same purpose, and discuss the limits of the approach.