Methods to extract molecular and bulk chemical information from series of complex mass spectra with limited mass resolution

Abstract

The resolution of mass spectrometers is often insufficient to conclusively identify all peaks that may be present in recorded spectra. Here, we present new methods to extract consistent molecular and bulk level chemical information by constrained fitting of series of complex organic mass spectra with multiple overlapping peaks. Possible individual peaks in a group of overlapping peaks are identified by both defining a chemical space and by free peak fitting. If simply all possible formulas from the chemical space would be used to fit each peak, the result would not be well constrained. The free peak fitting algorithm provides information about likely peak locations. A new algorithm then reconciles the results of both methods and produces a final peak list for use in subsequent fitting, while using all available experimental constraints. Comparison to ultra-high resolution data suggests that the real peak density is substantially higher than can be resolved with the instrument resolution. Bulk chemical properties such as carbon number (nC) and carbon oxidation state (OSC) can be calculated from the fit results. For mixtures of compounds dominated by C, H, O and N, bulk properties can be reliably extracted, even though some formula assignments may remain uncertain. This ability to retrieve correct bulk parameters even if not all assigned formulas are correct originates from the relationship between mass defects of individual peaks and the chemical parameters under our CHON composition assumptions. Retrieving consistent bulk parameters across series of many mass spectra is essential for extracting time trends, e.g. for field measurements taking place over several weeks. We illustrate the fitting method using a sample data set from a chemical ionization mass spectrometer with a resolution of approximately 4000 (M/dM), operated using acetate reagent ions. Spectral simulation experiments validate the analysis method by showing good agreement of intensities for many specific ions, as well as for bulk chemical parameters. An alternative method to directly extract bulk chemical information from the raw spectra without the need of any peak assignment or peak fitting is also introduced, which shows good agreement with the peak fitting results. The latter method can be applied very rapidly without the need for complex analysis procedures, e.g. as a quick online diagnostic during data acquisition.