Modelling protein solutions in membrane separation processes - from simple patchy colloids to realistic representation of proteins

We aim to create and validate a novel coarse-grained modelling approach for predicting acid-base equilibria in protein solutions separated from a buffer by a semi-permeable membrane. Such predictive modelling is crucial for optimising industrial separation processes like dialysis or ultra/dia-filtration. Models based on the mean-field approximation, commonly used in the industry, sometimes fail to predict the charge on proteins and the pH in the protein solution because they do not account for the irregular shapes of proteins, their charge distribution and protein-protein interactions. By accounting for these factors, our simulations will not only improve the prediction of pH, protein charge and ion partitioning but also enable predicting phase separation and precipitation as a function of pH, which is impossible with current models. We will validate our modelling using dialysis experiments on wellknown proteins. If successful, our approach will enhance the fundamental physico-chemical understanding needed to improve optimization of industrial processes.