Statistical methods for interpretation of high resolution mass spectra

Mass spectrometry is a powerful analytical technique used to characterize various compounds. A mass spectrum is a graph of ion intensity as a function of mass-to-charge ratio.; Protein study experiments generate thousands of mass spectra, generating an overload of data that necessitates the development of sophisticated data analysis methods. Our work aims at developing the following methods that allow for extraction of biochemically relevant information from mass spectra.; The maximum likelihood estimator together with the non-random parameter estimation method has been used to derive the mathematical relationship between the number of ions generated in a mass spectrometry experiment and the variance in the experimental isotopic distribution in a spectrum. Performance analysis of the method has been carried out using simulated and experimental data. The method can show a factor of two improvement over a previously developed method, and is applicable for any isotopically resolved spectrum.; A theoretical framework has been developed and tested against experiments for estimating high-precision elemental isotopic abundances from the experimental isotopic distributions. Higher molecular weights are particularly useful for a better estimate because the higher number of carbon atoms and isotopic peaks observed lead to a greater amount of information. This method circumvents some of the limitations experienced by the traditional isotope ratio mass spectrometry.; Charge state determination requires methods to accurately estimate the m/z difference between adjacent isotopic peaks. A new method for charge state determination using the Matched Filter approach has been developed and compared with the established methods under various conditions. Matched Filter method performs significantly better than the existing methods and has a particular advantage in cases involving overlapping isotopic distributions and low signal-to-noise ratio cases.; Algorithms have been developed and integrated as MasSPIKE (Mass Spectrum Interpretation and Kernel Extraction) for isotopic cluster identification, charge state determination, resolving overlapping isotopic distributions, alignment of the experimental isotopic distribution with the theoretical isotopic distribution, and reducing the isotopically resolved mass spectrum to a monoisotopic mass list. MasSPIKE has been used to characterize post-translational modifications for biologically interesting proteins Hemoglobin and H-Ras, allowing for differentiation of blood samples of diseased and healthy persons.