Prediction of protein retention in ion-exchange systems using molecular descriptors obtained from crystal structure

In this paper, a novel approach is described for the a priori prediction of protein retention in ion exchange systems. Quantitative structure retentio n relationship (QSRR) models based on a genetic algorithm/partial least squ ares approach were developed using experimental chromatographic data in con cert with molecular descriptors computed using protein crystal structures. The resulting QSRR models were well-correlated, with cross-validated r(2) v alues of 0.938 and 0.907, and the predictive power of these models was demo nstrated using proteins not included in the derivation of the models. Impor tantly, these models were able to predict selectivity reversals observed wi th two different stationary phase materials. To our knowledge, this is the first published example of predictive QSRR models of protein retention base d on crystal structure data.