An Integrated Experimental and Computational Approach to Proteomics: Scaling from High Resolution Qualitative Analysis to Quantitative Measurements with Confidence Evaluation

As a component of systems biology, proteomics aims to characterize the entire protein complement of an organism, including qualitative identification of protein types and quantitative measurement of protein abundance changes as a function of different cellular states. This dissertation presents an integrated experimental and computational approach to improve proteomic measurements, including qualitative measurements using Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR-MS) and quantitative measurements with statistically derived confidence evaluation. Although FT-ICR-MS provides high-performance mass measurements, its potential has not yet been fully explored for proteomics applications. A novel tandem mass spectrometry method was developed for FT-ICR-MS to obtain sequence tag information directly from intact proteins in a mixture. The interpretation of FT-ICR tandem mass spectra for sequence tagging was facilitated with a new graph-theoretical algorithm for separation of y- and b-ions. To scale FT-ICR-MS for general proteomic characterizations, low flow-rate liquid chromatography was integrated with FT-ICR-MS. The high- performance MS greatly enhanced the depth and quality of the proteomics measurements. In total, these studies demonstrated that FT-ICR-MS is of practical value for proteomic measurements, and that additional experimental and computational developments could make this into a robust and automated approach. Quantitative proteomics based on stable isotope labeling enables global gene expression profiling at the protein level. However, major challenges remain for extracting reliable protein quantification information from noisy mass spectrometric data. A principal component analysis algorithm was developed to accurately estimate peptide abundance ratios and to provide rigorous scores for their estimation variability and bias. The peptide quantification results were then processed by a novel profile likelihood algorithm to estimate protein abundance ratios with confidence interval evaluation. These algorithms were integrated into a computer program, ProRata, for automated data analysis. Quantitative proteomic measurements were conducted using ProRata, and integrated with transcriptomic analysis to study the anaerobic metabolism of p-coumarate in Rhodopseudomonas palustris. This study yielded a putative cellular pathway for p-coumarate catabolism. In the research described here, a substantial advancement in both qualitative and quantitative proteomic measurements was achieved using an integrated experiment and computational approach. The improved proteomic measurements can help elucidate a range of biological processes