One of the most puzzling aspects of modern computational chemistry is our continuing inability to predict the gross structure of biological macromolecules, especially proteins, from sequence [1]. This is puzzling because we can easily, with convincing accuracy, predict the microstructural features such as bond lengths, angles and conformation of individual sidechains. It is difficult to conceive of analogous situations in other fields of science; more often than not the "macro" is understood and manipulated much more readily than the "micro".There are, of course, a multitude of reasons for this computational impotence. The complexity of biological systems is probably the most obvious. For a single protein we have to consider millions of potential contacts, thousands of degrees of freedom and likely hundreds of individual molecules if water is part of the model. The modeling of water with current methodology is difficult. Where should water be considered as bulk, which is conventionally dismissed by using a non-unity dielectric, and where should individual (ordered and disordered) water molecules be explicitly modeled? It is obvious that water is crucial; not only because of its role as a potent hydrogen bond mediator that supports protein structure, but also because significant entropy arises from displacement and/or movement of water in and out of the system [2]. The lack of adequate representation for entropy, despite the accuracy of the calculations for enthalpy, compromises our understanding of any real biomacromolecular structure or process. Complex dynamics simulations run over significant time frames would seem to be a reasonable response to these problems, but application continues to be expensive; and, even when the starting point of a dynamics simulation is the known crystallographic structure, after significant simulation time periods the final structure often loses much of the biomacromolecule's original secondary and tertiary structure.For some time we have been interested in understanding and exploiting biological structure for a variety of purposes related to developing new disease treatments. The processes of ligand binding, protein-protein associations, protein-DNA associations and etc. are all fundamental to the design of new therapeutic agents. Our approach has been to develop a simple computational model that is based not on force fields, but on an actual experimental free energy measurement -the partition coefficient for octanol/water solubility, logP o/w . This parameter has been used for several decades as a measure of lipid transport that can be often related to the in vivo biological activity of drug candidate molecules [3,4]. LogP o/w has also been shown in many cases to correlate with ligand binding measured in vitro [5,6]. At its simplest level logP o/w and its component fragment and atom terms reveal the type of interactions that the molecule/fragment/atom are able to make with another species. LogP o/w (or fragment/atom constants) less than zero is indicative of a polar (hydrophilic...