Mass spectrometry (MS) based proteomics has emerged as a key technology in the search for disease- associated biomarkers. State-of-the-art instruments can identify thousands of proteins in a single sample by 'shotgun'proteomic analysis, where protein mixtures are proteolyzed into peptides, separated by one or more chromatographic steps, and analyzed by peptide dissociation using tandem mass spectrometry (MS/MS). The goal of this approach is to create new technologies for the accurate detection of proteins within complex samples. Achieving this target is currently limited by the major problem of inferring the peptide sequence from MS/MS spectra by sequence database searching: spectra are compared to "model spectra" generated from database sequences. Current algorithms suffer from poor accuracy and discrimination due to the use of simple models for predicting spectra, which ignores the rich information contained in the relative intensities of peaks in a typical MS/MS. Consequently, there is a vital need for more accurate models to predict MS/MS spectrum intensities from peptide sequences. In this proposal, we will develop a new and innovative kinetic model for predicting peptide fragmentation MS/MS spectra, and use the model to develop MS/MS identification algorithms with high discrimatory power. Spectra simulated by the kinetic model will then be used to design selected reaction monitoring (SRM) assays, which have become a critically important technique for measuring targeted sets of proteins in human biomarker studies. This will solve a bottleneck for widespread adoption of SRM methods for biomarker discovery, which is currently hindered by the slow process of identifying and optimizing SRM transitions for the assays. The following specific aims are (1) Develop an optimized kinetic model of gas-phase peptide fragmentation which predicts MS/MS spectra for any peptide sequence. Model parameters will be fit using the Levenberg- Marquardt algorithm, a robust method for non-linear least squares. (2) Extend the model to predict MS/MS fragmentation of phosphopeptides. The approaches developed in this aim can be extended to other disease- relevant post-translational modifications which profoundly alter peptide fragmentation and interfere with MS/MS identification. (3) Develop a route to successful implementation of spectrum-to-spectrum matching algorithms, an entirely new approach for large scale identification of proteins, in which MS/MS are searched directly against libraries of predicted spectra, simulated using our prototype kinetic model. We use predicted spectra to bypass the need for sequence databases, and spectrum-to-sequence strategies altogether. (4) Develop an algorithm for de novo prediction of selected reaction monitoring (SRM) assays for highly multiplexed quantitative measurement of proteins in complex mixtures.
Mass spectrometry-based proteomics has emerged as a key technology in the search for useful protein biomarkers, and holds many promises for early detection of disease, prediction of drug efficacy and resistance, and targeted molecular therapies. The field is currently limited by the major problem of inferring the peptide sequence from a fragmentation mass spectrum - until this problem is solved, many potential applications of proteomics to human health will not be achieved. We will develop a kinetic model to predict peptide fragmentation spectra for any peptide sequence;a method that will enable comprehensive protein profiling in human biofluids, and the rapid design of selected reaction monitoring (SRM) assays, which have become a critically important technique for measuring targeted sets of proteins in human biomarker studies.