Studies On Key Issues Of Mass Spectrometry Data Interpreting In Proteomics

Mass spectrometry(MS) has become the core technology of proteomics research. MS data is the fundamental resource for proteome information mining, and MS data processing is the core of computational proteomics. It remains a large challenge for proteome research to interpret complex peptide information from simple physical signals, and then to extend to address specific biological problems. In this dissertation, we complete the following work around the mass spectrometry data interpreting in proteomics:(1) Basic characteristics of the peptide isotope peak are studied. The isotope peaks of peptide precursor ions are present in the mass spectra. To better understand and interpret the MS data, the basic properties of individual isotopes and isotope peak clusters are studied. In this paper, the high resolution MS data is selected as the research object. Through a lot of exploratory analysis, we find out the basic characteristics of peptide isotopic peaks, such as the ion sampling, peak width and peak type, and the composition and abundance index of the isotope peak clusters, which has laid a good foundation for further analyzing the MS data. Main results are summarized as follows: multiple quadratic curve distributions are found between the peak width and m/z of the isotope peak on given MS platform; the relationship between the ion sampling interval and its m/z is repeatable on the same MS platform, but not on different MS platforms; regardless of all ions or the multiple parent ions with same m/z, the distribution of the sampling points are found close to the Gaussian distribution; for the multiple parent ions with same m/z, the number of the sampling points is positively correlated with the peak intensity and is weak negatively correlated with RT; the number of the compositive peaks showed positive correlations with the peptide abundance, molecular weight and RT; the distribution of isotope peaks represented by their respective maximum is closest to the theoretical distribution.(2) The abundance measurement errors of peptide isotope peaks are studied. The isotope abundance measurement accuracy is an important indicator for the performance of a mass spectrometer. The research on the isotope abundance measurement errors is helpful to reveal the remaining problem in the output process of MS data, thus providing clues for improving existing data analysis methods and instrument performance. In this thesis, the peak maximum is chosed as the quantitative indicator for isotope peaks. From different perspective, the abundance error distributions of peptide isotopic peaks are studied. We have figured out the isotopic peak abundance error characteristics including the normalized error and local average error of single isotope peaks, the total error of all isotopic peaks, and the correlations between different isotopic peak abundance errors. Main results are summarized as follows: compared with the additive error model and the multiple error model, the normalized error model is found with the highest sensitivity; the random error is reduced and systematic error is highlighted through computing the local average error at ion chromatographic peak level; the distribution of all peptide isotope peak errors can be fitted perfectly by a mixed Gaussian distribution with two kernel functions; the correlations between the abundance errors of peptide isotope peaks with different index are found in varying degrees; the correlations and systematic errors may be caused by the suppression of high abundance peaks on low abundance peaks.(3) The correction and application of peptide isotope abundance errors are studied. As the isotope peak abundance of peptide represent a peptide quantitative information, so the peptide quantification accuracy is affected inevitably by the measurement errors of the isotope peak abundance. In this thesis, two isotope abundance error correction algorithms are proposed from different perspective, namely the error correction algorithm based on the normalized ratios and the iterative correction algorithm based on multiple linear regression. Two correction algorithms are evaluated using multiple experimental datasets. The results are obtained as follows: the error correction algorithm based on the normalized ratios can eliminate systematic errors, but limited in improving the quantitative performance; the iterative correction algorithm based on multiple linear regression is worse than the former in terms of reducing the systematic errors, but more effective in improving the quantitative performance.(4) An improved workflow for identifying ubiquitin(Ub) / ubiquitin-like protein(Ubl) modification sites is proposed. Trypic digestion of many of Ubls(e.g. SUMO-1,-2,-3, FAT10) or Lys C digestion of Ub, NEDD8 or ISG15 results in a long side chain left on the target lysine residues, resulting in a challenge to unbaisedly and sensitively identify Ub and Ubls conjugation sites by high-throughput mass spectrometry. Firstly, we analyze the deficiencies of the existing method. Secondly, an improved workflow for identifying Ub/Ubl modification sites is proposed to overcome its shortcomings. Thirdly, the new workflow is evaluated using two public datasets. Compared to the original Chop NSpice method, the improved workflow has three advantages:(1) the sensitivity of identification of Ub and Ubl conjugation sites can be increased using Ubl Search;(2) false positive results and the time required for results validation can be both reduced;(3) When more than one lysine residues exist in the target peptide sequence except the peptide C-terminal lysine residue, which lysine residue is the conjugation site can be automatically identified. Thus,the improved workflow is particularly sutiable for processing large-scale mass spectra datasets from high-throughput experiments.(5) The FAT10 modification sites are firstly identified using the improved workflow. Cooperation with Beijing Proteome Research Center, the proposed workflow is used to analyze large-scale FAT10 MS data. As a result, the FAT10 modification sites are firstly identified in the world. Examination of a six amino acid window adjacent to modified lysines reveals that the modified lysines have a slight trendency to be localized in regions enriched in hydrophilic residues. Compared with the SUMO consensus sequence, only the second amino acid adjacent to the modified lysine has a same conservation. In total, 175 FAT10 target proteins are identified using standard database search and label-free quantitative technique. The FAT10 target proteins participate in various biological processes, such as protein transport, protein folding, RNA processing and macromolecular complex assembly, post-transcriptional regulation of gene expression, protein localization, regulation of apoptosis and Golgi vesicles budding.