Optimal Decharging and Clustering of Charge Ladders Generated in ESI−MS

Abstract

In electrospray ionization mass spectrometry (ESI−MS), peptide and protein ions are usually observed in multiple charge states. Moreover, adduction of the multiply charged species with other ions frequently results in quite complex signal patterns for a single analyte, which significantly complicates the derivation of quantitative information from the mass spectra. Labeling strategies targeting the MS1 level further aggravate this situation, as multiple biological states such as healthy or diseased must be represented simultaneously. We developed an integer linear programming (ILP) approach, which can cluster signals belonging to the same peptide or protein. The algorithm is general in that it models all possible shifts of signals along the m/z axis. These shifts can be induced by different charge states of the compound, the presence of adducts (e.g., potassium or sodium), and/or a fixed mass label (e.g., from ICAT or nicotinic acid labeling), or any combination of the above. We show that our approach can be used to infer more features in labeled data sets, correct wrong charge assignments even in high-resolution MS, improve mass precision, and cluster charged species in different charge states and several adduct types.