Mobility modeling of peptides in capillary electrophoresis

Recent rapid developments in proteomics require high-resolution separation of a large number of peptides for their downstream identification by mass spectrometry. Capillary electrophoresis (CE) is an electric-field-mediated bioanalytical technique capable of rapid, high-resolution separation of very complex sample mixtures. Development of CE methods for adequate separation of a large number of peptides is usually a time-consuming task.Application of model-based approaches to predict peptide mobilities in CE from known physicochemical properties can shorten tedious experimental optimization of separation. This endeavor requires specification of structural descriptors followed by selection of appropriate modeling methods. To date, numerous theoretical predictive models have been developed, mostly based on Stokes’ Law to relate peptide mobilities to structural properties (e.g., charge and size). However, these two-variable models could not successfully predict electrophoretic mobilities for all categories of peptides with a reasonable degree of accuracy.To address the shortcomings of the two-variable models, new strategies were recently introduced, including the usage of additional peptide descriptors or applying non-linear modeling (e.g., artificial neural networks), to attain more accurate, robust prediction. Effective application of machine-learning techniques to the development of predictive models has consolidated conjecture on non-linear relationships between peptide structural descriptors and their electrophoretic mobilities.In this article, we review recent advances in CE mobility modeling of peptides, particularly in respect to predicting optimal separation conditions for the analysis of highly complex peptide mixtures in proteomics applications.