Mean force, potentials of

We also assume that the total potential energy is pairwise additive. Hence, we write 

Note that the integration in the numerator of (2.119) is over R3.RN, and the quantity V U(R, R ) is independent of these variables. We also introduce the conditional probability density of finding the N— 2 particles at a specified configuration R3. RN given that the two particles are at R and R , namely 

We can further simplify (2.121) by noting that the sum over i produces N— 2 equal terms, i.e., 

The quantity p(R/R, R ), introduced in (2.122), is the conditional density at a point R, given two particles at R1 and R . This is a straightforward generalization of the conditional density introduced in section 2.4. The total force acting on 1 can now be written as 

This form is useful in the study of forces applied to solutes or to groups in proteins, in aqueous solutions. The first term is referred to as the direct force and the second term as the solvent-induced force. 

---

The pair correlation fiinction has a simple physical interpretation as the potential of mean force between two particles separated by a distance r... 

The concept of the potential of mean force can be extended to mixtures and solutions. Consider two ions in a sea of water molecules at fixed temperature T and solvent density p. The potential of mean force w r) is the direct interaction between the ions u.j r) = plus the interaction between the ions tln-ough water... 

McMillan-Mayer theory of solutions [1,2], which essentially seeks to partition the interaction potential into tln-ee parts that due to the interaction between the solvent molecules themselves, that due to die interaction between the solvent and the solute and that due to the interaction between the solute molecules dispersed within the solvent. The main difference from the dilute fluid results presented above is that the potential energy u(r.p is replaced by the potential of mean force W(rp for two particles and, for particles of solute in the solvent, by the expression... 

The McMillan-Mayer theory allows us to develop a fomialism similar to that of a dilute interacting fluid for solute dispersed in the solvent provided that a sensible description of W can be given. At the Ihnit of dilution, when intersolute interactions can be neglected, we know that the chemical potential of a can be written as = W (a s) + IcT In where W(a s) is the potential of mean force for the interaction of a solute... 

At low solvent density, where isolated binary collisions prevail, the radial distribution fiinction g(r) is simply related to the pair potential u(r) via g ir) = exp[-n(r)//r7]. Correspondingly, at higher density one defines a fiinction w r) = -kT a[g r). It can be shown that the gradient of this fiinction is equivalent to the mean force between two particles obtamed by holding them at fixed distance r and averaging over the remaining N -2 particles of the system. Hence w r) is called the potential of mean force. Choosing the low-density system as a reference state one has the relation... 

The simple difhision model of the cage effect again can be improved by taking effects of the local solvent structure, i.e. hydrodynamic repulsion, into account in the same way as discussed above for bimolecular reactions. The consequence is that the potential of mean force tends to favour escape at larger distances > 1,5R) more than it enliances caging at small distances, leading to larger overall photodissociation quantum yields [H6, 117]. 

Having separated the dynamical from equilibrium (or, more accurately, quasi-equilibrium) effects, one can readily discover the origin of the activation free energy and define the concept of the potential of mean force by analysis of the expression for the TST rate constant, k in (A3.8.3). The latter can be written as [7]... 

Haynes G R and Voth G A 1993 The dependence of the potential of mean force on the solvent friction consequences for condensed phase activated rate theories J. Chem. Phys. 99 8005... 

Colloidal particles can be seen as large, model atoms . In what follows we assume that particles with a typical radius <3 = lOO nm are studied, about lO times as large as atoms. Usually, the solvent is considered to be a homogeneous medium, characterized by bulk properties such as the density p and dielectric constant t. A full statistical mechanical description of the system would involve all colloid and solvent degrees of freedom, which tend to be intractable. Instead, the potential of mean force, V, is used, in which the interactions between colloidal particles are averaged over... 

Chemical reaction dynamics is an attempt to understand chemical reactions at tire level of individual quantum states. Much work has been done on isolated molecules in molecular beams, but it is unlikely tliat tliis infonnation can be used to understand condensed phase chemistry at tire same level [8]. In a batli, tire reacting solute s potential energy surface is altered by botli dynamic and static effects. The static effect is characterized by a potential of mean force. The dynamical effects are characterized by tire force-correlation fimction or tire frequency-dependent friction [8]. 

The first requirement is the definition of a low-dimensional space of reaction coordinates that still captures the essential dynamics of the processes we consider. Motions in the perpendicular null space should have irrelevant detail and equilibrate fast, preferably on a time scale that is separated from the time scale of the essential motions. Motions in the two spaces are separated much like is done in the Born-Oppenheimer approximation. The average influence of the fast motions on the essential degrees of freedom must be taken into account this concerns (i) correlations with positions expressed in a potential of mean force, (ii) correlations with velocities expressed in frictional terms, and iit) an uncorrelated remainder that can be modeled by stochastic terms. Of course, this scheme is the general idea behind the well-known Langevin and Brownian dynamics. 

However, in many applications the essential space cannot be reduced to only one degree of freedom, and the statistics of the force fluctuation or of the spatial distribution may appear to be too poor to allow for an accurate determination of a multidimensional potential of mean force. An example is the potential of mean force between two ions in aqueous solution the momentaneous forces are two orders of magnitude larger than their average which means that an error of 1% in the average requires a simulation length of 10 times the correlation time of the fluctuating force. This is in practice prohibitive. The errors do not result from incorrect force fields, but they are of a statistical nature even an exact force field would not suffice. 

Thus one must rely on macroscopic theories and empirical adjustments for the determination of potentials of mean force. Such empirical adjustments use free energy data as solubilities, partition coefficients, virial coefficients, phase diagrams, etc., while the frictional terms are derived from diffusion coefficients and macroscopic theories for hydrodynamic interactions. In this whole field of enquiry progress is slow and much work (and thought ) will be needed in the future. 

Besides yielding qualitative information, these biologically and pharmaceutically motivated applications of SMD can also yield quantitative information about the binding potential of the ligand-receptor complex. A first advance in the reconstruction of the thermodynamic potential from SMD data by discounting irreversible work was made by Balsera et al. (1997) as outlined in Sect. Reconstruction of the potential of mean force below. 

In this section we describe the behavior of a ligand subjected to three types of external forces a constant force, forces exerted by a moving stiff harmonic spring, and forces exerted by a soft harmonic spring. We then present a method of reconstruction of the potential of mean force from SMD force measurements employing a stiff spring (Izrailev et al., 1997 Balsera ct al., 1997). 

Such a free energy is called a potential of mean force. Average values of Fs can be computed in dynamics simulations (which sample a Boltzmann distribution), and the integral can be estimated from a series of calculations at several values of s. A third method computes the free energy for perturbing the system by a finite step in s, for example, from si to S2, with... 

The problems that occur when one tries to estimate affinity in terms of component terms do not arise when perturbation methods are used with simulations in order to compute potentials of mean force or free energies for molecular transformations simulations use a simple physical force field and thereby implicitly include all component terms discussed earlier. We have used the molecular transformation approach to compute binding affinities from these first principles [14]. The basic approach had been introduced in early work, in which we studied the affinity of xenon for myoglobin [11]. The procedure was to gradually decrease the interactions between xenon atom and protein, and compute the free energy change by standard perturbation methods, cf. (10). An (issential component is to impose a restraint on the... 

The free energy profile or potential of mean force along a conformational coordinate may be defined as... 

The additional integration in the final equation is similar to integration of potential of mean force. Equation (29) is more difficult to compute since Xq is a multidimensional vector. 

One approach to this problem is to use a potential of mean force (PMF), which describes he the free energy changes as a particular coordinate (such as the separation of two atoms or t torsion angle of a bond) is varied. The free energy change described by the potential of me force includes the averaged effects of the solvent. 

A simulation performed using a potential of mean force enables the modulating effects of t solvent to be taken into account. The solvent also influences the dynamic beliaviour of t... 

---