Biomolecules typically exist in aqueous environment. The influence of water needs to be taken into account in all theoretical studies concerning living matter at atomistic level. Computer simulations provide means for doing so, however, approaches involving water in atomic representation are often prohibitively expensive, while the existing simplified, so called implicit solvent models are only moderately successful.
We will present a novel method for modeling of biomolecular hydration. Based on discrete solvent representation and mean field approach, the method is capable of addressing deficiencies of the implicit solvent models, while maintaining their computational efficiency. We will demonstrate that the proposed model correctly reproduces experimental hydration free energies for an extensive set of roughly 700 diverse organic compounds, and accurately predicts the distribution of water molecules buried within protein structures. Those capabilities make it a useful tool for computational and structural biology, in particular as an aid for experimental methods aimed at the prediction of macromolecular structures such as X-ray crystallography or NMR, as well as in application oriented areas such as computer aided drug design.