Fluctuation theory of solutions, introduced by Kirkwood and Buff in 1951, relates particle number fluctuations of small scales to the global thermodynamic properties of the system. Cosolvent effects on (bio)solutes in aqueous solutions can be modeled in molecular dynamics simulations and the fluctuation theory of solutions, termed as Kirkwood-Buff (KB) theory, provides a theoretical route to analyze the specific interactions between the solute and the cosolvent in terms of the preferential solvation or preferential binding which is important to study, amongst others, the conformational changes in biomacromolecules under different cosolvent conditions. The KB theory can be used to relate the simulation data on the integrals of the pair-correlation functions between the system components over volume to the experimentally observable thermodynamic quantities such as change in the chemical potential of the solutes with varying cosolvent concentrations. Although computer simulations provide the interaction mechanisms or the dynamics of a system at molecular level, but they are confronted with many challenges related to the limitations in computational power, accuracy of the models and straight-forward comparison with experimental data. The number of particles in a system can be reduced significantly by grouping several particles in single interaction-sites, termed as coarse-grained beads, which leads to significant speed-up of the simulations. With coarse-graining, simulations of larger systems with longer time-scale are possible which are required for the most of the biological processes. In this thesis we use the KB theory to develop simplified coarse-grained models for aqueous binary and ternary mixtures. On the other hand, by resolving the integrals of the pair-correlation functions to the contributions arising from different modes of spatial separation between the solution components, the KB theory is also used to explain the ion-specific pairing mechanisms between Hofmeister ions and the ion-specific changes in the solvation thermodynamics of solutes in aqueous solutions with all-atom simulations.

This thesis includes a theoretical account on the Kirkwood-Buff theory explaining the relevant thermodynamic relations which is followed by a review on the applications of the Kirkwood-Buff theory to the computer simulations of aqueous solutions. Then this thesis proposes a new method of coarse-graining by combining the structure-based Iterative Boltzmann Inversion (IBI) method and the Kirkwood-Buff (KB) theory. The method, KB-IBI, is applied to binary mixtures of urea-water and benzene-water and the single-site coarse-grained potentials for the molecules are found to be consistent with the atomistic pair-correlations and the variations in the urea or benzene chemical potentials with different solution concentrations. As urea serves as a chemical denaturant for proteins, application of these coarse-grained potentials to the ternary mixtures of solutes in urea-water would be the first step towards modeling the urea-driven conformational changes in biomolecules. So the preferential interactions between benzene and urea are studied with single-site coarse-grained models and the variation in the solvation free-energy of benzene with different urea concentrations has been reproduced in agreement with the atomistic model. The representability and the convergence of the KB-IBI coarse-grained models at a particular state-point where the model is parametrized are discussed in terms of the thermodynamic quantities such as pressure, potential energy and the variation in the solvation free-energy for the systems of pure water, binary urea-water mixture and ternary benzene-urea-water mixtures at infinite benzene dilution. The transferability issue of the KB-IBI potentials at different urea concentrations has also been examined and a cluster analysis of benzene in urea-water solutions is discussed.

With all-atomistic simulations the application of the KB theory in an analysis of monovalent alkali cation pairing with biologically relevant anions such as acetate or phosphate has revealed a ion-specific variation in the water-mediated ion-pairs which leads to the variation in the activity of the salts. Contributions to the integrals of the pair-correlation functions originating from the different ion pairing modes, namely contact ion-pairs (direct pairing between the cation and the anion), solvent-shared ion-pairs (solvation-shells of the ions are shared) or solvent-separated ion-pairs (solvation-shells of the ions are separated), have been analysed. It has been found that solvent-separated ion-pairing mechanism for phosphate and solvent-shared mechanism for acetate play the major role in the ion-specific changes in the salt activity in the solution; whereas for chloride solutions contact ion-pairing mechanism prevails over solvent-mediated mechanisms. For the ternary systems of solutes in the salt-solutions, the interactions between benzene and the ions in aqueous solutions of the alkali chlorides have been studied with KB theory and different force-fields models have been tested. Simulation data suggest that the direct correlations between benzene and ions play more significant role rather than the indirect ion-pairing to explain the ion-specific decrease in the solubility of benzene, termed as ion-specific salting-out of benzene, upon addition of salts. A geometric packing of hydrated lithium ions around benzene is found to be the reason of lithium chloride being less salting-out agent than sodium chloride or potassium chloride. Calculation of the integrals of the pair-correlation functions over volume, termed as Kirkwood-Buff integrals (KBIs) and which are the key quantities in the KB theory to relate the local pair-structures to the thermodynamic quantities, does come with many technical issues. Calculation of more precise KBIs and the effect of the system size and the simulation time on the KBIs are discussed with the help of binary mixtures of urea-water and methanol-water where the convergence issues of the KBIs are more pronounced due to the microheterogeneity of the solutions and slower dynamics of the local domains.

This thesis serves as an account on the diverse applicability of KB theory to the computer simulations of biologically important systems. The newly developed coarse-graining method, KB-IBI, can be thought as a novel step towards modeling aqueous single-phase solutions and can potentially be extended to model polymers or biomacromolecules in urea-water or other cosolvent-water solutions. Also the KB theory in general can be used to quantify the preferential solvation of the solutes with cosolvents and to study the conformational changes in the biomolecules, provided that the technical issues in the calculation of the KBIs are addressed properly.