The number of bioprocess design alternatives grows dramatically when possible bioseparation methods are considered. Separation methods exploit different driving forces (gravitational, centrifugal, chemical potential, electric potential and pressure differences) that are rooted in the differences in molecular properties between the target compound and its contaminants. Industrial practice often involves intuitive qualitative concepts based on substantial experimental effort, creating many suboptimal situations. Hence, there is a need for thermodynamic data and/or predictive models, requiring an absolute minimum of experimental effort. We have developed several generalized models for predicting thermodynamic properties such as solubilities, activity coefficients, chemical potentials, and partition coefficients in solvent mixtures of small biomolecules from a few measurable thermodynamic parameters. These models allow translation of measurable properties, such as the second osmotic virial coefficient, B2, among different bioseparation techniques, facilitating rapid screening and design of unit operations. However, larger species such as proteins are more complicated because their phase behavior may be strongly affected by conformational changes. This paper extends the models to proteins and protein mixtures. Combined with high-throughput experimentation for rapid generation of thermodynamic data such as for B2, a framework and methodology will be created for systematic and faster bioprocess design.