adaptive biasing force
Overcoming free energy barriers in molecular simulations using an average force

Historical background

Theoretical foundations
Success stories
Contact us

How to cite ABF

The references for the ABF method are Darve and Pohorille, J. Chem. Phys. 2001 and Darve, Rodriguez-Gomez and Pohorille, J. Chem. Phys. 2008.
The reference for the new NAMD formulation and implementation is Hénin, Fiorin, Chipot, and Klein, J. Chem. Theory Comput. 2010.


Calculating free energies using average force

Eric Darve and Andrew Pohorille

A new, general formula that connects the derivatives of the free energy along the selected, generalized coordinates of the system with the instantaneous force acting on these coordinates is derived. The instantaneous force is defined as the force acting on the coordinate of interest so that when it is subtracted from the equations of motion the acceleration along this coordinate is zero. The formula applies to simulations in which the selected coordinates are either unconstrained or constrained to fixed values. It is shown that in the latter case the formula reduces to the expression previously derived by den Otter and Briels [Mol. Phys. 98, 773 (2000)]. If simulations are carried out without constraining the coordinates of interest, the formula leads to a new method for calculating the free energy changes along these coordinates. This method is tested in two examples — rotation around the C–C bond of 1,2-dichloroethane immersed in water and transfer of fluoromethane across the water-hexane interface. The calculated free energies are compared with those obtained by two commonly used methods. One of them relies on determining the probability density function of finding the system at different values of the selected coordinate and the other requires calculating the average force at discrete locations along this coordinate in a series of constrained simulations. The free energies calculated by these three methods are in excellent agreement. The relative advantages of each method are discussed.
J. Chem. Phys.
2001, 115, 9169-9183.

Calculating free energies using a scaled-force molecular dynamics algorithm

Eric Darve, Michael A. Wilson and Andrew Pohorille

We propose and test a family of methods to calculate the free energy along a generalized coordinate, ξ, based on computing the force acting on this coordinate. First, we derive a formula that connects the free energy in unconstrained simulations with the force of constraint that can be readily calculated numerically. Then, we consider two methods, which improve the efficiency of the free energy calculation by yielding uniform or nearly uniform sampling of ξ. Both rely on modifying the force acting on ξ. In one method, this force is replaced by a force with zero mean and ξ is advanced quasistatically. In the second method, the force is augmented adaptively by a biasing force. We provide formulas for calculating the free energy of the unmodified system from the forces acting in these modified, non-Hamiltonian systems. Using conformational transitions in 1,2-dichloroethane as a test case, we show that both methods perform very well.
Mol. Sim.
200128, 113-144.

Overcoming free energy barriers using unconstrained molecular dynamics simulations

Jérôme Hénin and Christophe Chipot

Association of unconstrained molecular dynamics (MD) and the formalisms of thermodynamic integration and average force [Darve and Pohorille, J. Chem. Phys. 115, 9169 (2001)] have been employed to determine potentials of mean force. When implemented in a general MD code, the additional computational effort, compared to other standard, unconstrained simulations, is marginal. The force acting along a chosen reaction coordinate ξ is estimated from the individual forces exerted on the chemical system and accumulated as the simulation progresses. The estimated free energy derivative computed for small intervals of ξ is canceled by an adaptive bias to overcome the barriers of the free energy landscape. Evolution of the system along the reaction coordinate is, thus, limited by its sole self-diffusion properties. The illustrative examples of the reversible unfolding of deca-L-alanine, the association of acetate and guanidinium ions in water, the dimerization of methane in water, and its transfer across the water liquid-vapor interface are examined to probe the efficiency of the method.
J. Chem. Phys. 2004, 121, 2904-2914

Exploring the free-energy landscape of a short peptide using an average force

Christophe Chipot and Jérôme Hénin

The reversible folding of deca-alanine is chosen as a test case for characterizing a method that uses an adaptive biasing force (ABF) to escape from the minima and overcome the barriers of the free-energy landscape. This approach relies on the continuous estimation of a biasing force that yields a Hamiltonian in which no average force is exerted along the ordering parameter ξ. Optimizing the parameters that control how the ABF is applied, the method is shown to be extremely effective when a nonequivocal ordering parameter can be defined to explore the folding pathway of the peptide. Starting from a β-turn motif and restraining ξ to a region of the conformational space that extends from the α-helical state to an ensemble of extended structures, the ABF scheme is successful in folding the peptide chain into a compact alpha helix. Sampling of this conformation is, however, marginal when the range of ξ values embraces arrangements of greater compactness, hence demonstrating the inherent limitations of free-energy methods when ambiguous ordering parameters are utilized.
J. Chem. Phys.
2005, 123, 244906

Assessing the efficiency of free energy calculation methods

David Rodriguez-Gomez, Eric Darve and Andrew Pohorille

The efficiencies of two recently developed methods for calculating free energy changes along a generalized coordinate in a system are discussed in the context of other, related approaches. One method is based on Jarzynski's identity [Phys. Rev. Lett. 78, 2690 (1997)]. The second method relies on thermodynamic integration of the average force and is called the adaptive biasing force method [Darve and Pohorille, J. Chem. Phys. 115, 9169 (2001)]. Both methods are designed such that the system evolves along the chosen coordinate(s) without experiencing free energy barriers and they require calculating the instantaneous, unconstrained force acting on this coordinate using the formula derived by Darve and Pohorille. Efficiencies are analyzed by comparing analytical estimates of statistical errors and by considering two numerical examples—internal rotation of hydrated 1,2-dichloroethane and transfer of fluoromethane across a water-hexane interface. The efficiencies of both methods are approximately equal in the first but not in the second case. During transfer of fluoromethane the system is easily driven away from equilibrium and, therefore, the performance of the method based on Jarzynski's identity is poor.
J. Chem. Phys.
2004, 120, 3563-3578

Computation of free energy profiles with parallel adaptive dynamics

Tony Lelièvre, Mathias Rousset, and Gabriel Stoltz

We propose a formulation of an adaptive computation of free energy differences, in the adaptive biasing force or nonequilibrium metadynamics spirit, using conditional distributions of samples of configurations which evolve in time. This allows us to present a truly unifying framework for these methods, and to prove convergence results for certain classes of algorithms. From a numerical viewpoint, a parallel implementation of these methods is very natural, the replicas interacting through the reconstructed free energy. We demonstrate how to improve this parallel implementation by resorting to some selection mechanism on the replicas. This is illustrated by computations on a model system of conformational changes.
J. Chem. Phys. 2007, 126, 134111

Adaptive biasing force method for scalar and vector free energy calculations

Eric Darve, David Rodriguez-Gomez and Andrew Pohorille

In free energy calculations based on thermodynamic integration, it is necessary to compute the derivatives of the free energy as a function of one (scalar case) or several (vector case) order parameters. We derive in a compact way a general formulation for evaluating these derivatives as the average of a mean force acting on the order parameters, which involves first derivatives with respect to both Cartesian coordinates and time. This is in contrast with the previously derived formulas, which require first and second derivatives of the order parameter with respect to Cartesian coordinates. As illustrated in a concrete example, the main advantage of this new formulation is the simplicity of its use, especially for complicated order parameters. It is also straightforward to implement in a molecular dynamics code, as can be seen from the pseudocode given at the end. We further discuss how the approach based on time derivatives can be combined with the adaptive biasing force method, an enhanced sampling technique that rapidly yields uniform sampling of the order parameters, and by doing so greatly improves the efficiency of free energy calculations. Using the backbone dihedral angles Phi and Psi in N-acetylalanyl-N'-methylamide as a numerical example, we present a technique to reconstruct the free energy from its derivatives, a calculation that presents some difficulties in the vector case because of the statistical errors affecting the derivatives.
J. Chem. Phys.
2008, 128(14), 144120

3d FE landscape

Exploring multidimensional free energy landscapes using time-dependent biases on collective variables

Jérôme Hénin, Giacomo Fiorin,
Christophe Chipot and Michael L. Klein

A new implementation of the adaptive biasing force (ABF) method is described. This implementation supports a wide range of collective variables and can be applied to the computation of multidimensional energy profiles. It is provided to the community as part of a code that implements several analogous methods, including metadynamics. ABF and metadynamics have not previously been tested side by side on identical systems. Here, numerical tests are carried out on processes including conformational changes in model peptides and translocation of a halide ion across a lipid membrane through a peptide nanotube. On the basis of these examples, we discuss similarities and differences between the ABF and metadynamics schemes. Both approaches provide enhanced sampling and free energy profiles in quantitative agreement with each other in different applications. The method of choice depends on the dimension of the reaction coordinate space, the height of the barriers, and the relaxation times of degrees of freedom in the orthogonal space, which are not explicitly described by the chosen collective variables.
J. Chem. Theory Comput.
2010, 6(1), 35-47