An interdisciplinary team at Rensselaer Polytechnic Institute (RPI) in Troy, NY, has developed a computer program that uses protein structural information to predict how proteins separate on a chromatography column.1 "This is something that is usually done by trial and error, but in this case we've used information from the experiment to build a computer model that enables us to make that prediction," explains coauthor Curt Breneman, a computational chemist at RPI.
The technique first takes experimental structural and sequence data to predict thermodynamic isotherm parameters (including shape and electron density); these constraints are then used to predict the protein's behavior on a column. Coauthor Steven Cramer, a chemical engineer at RPI, explains that the model can be trained to predict the time a protein will come off a column, what the peak shape will look like, and what will happen if salt or concentration is changed. "It's the...