MS Informatics: Using Bioinformatic Tools to Enhance MS-based Neuropeptidomics and Proteomics

Nowadays, mass spectrometry (MS)-based proteomics is playing a leading role to enable deep investigation of cellular proteomes. However, the explosive amount of data generated by mass spectrometry makes it difficult to map these complex proteomics data to biological processes. This dissertation research focuses on the development of bioinformatic tools to accelerate and enhance MS-based neuropeptidomics and proteomics. Representing the largest group of signaling molecules, neuropeptides regulate many biological processes such as locomotion, feeding, learning and memory and etc. Characterization of neuropeptides is the first step towards understanding how the neural circuitry functions. An integrated analytical platform combining multi-faceted mass spectrometric approaches and in silico data mining techniques has been developed to discover neuropeptides in Callinectes sapidus and Carcinus maenas. Multiple ionization techniques coupled to various fractionation methods are used for neuronal tissue extract analysis and greatly improved the neuropeptidome coverage. Additionally, in silico data mining techniques including in silico transcriptomics and public database mining facilitate the discovery of previously known neuropeptides and large neuropeptides. A database is constructed for the storage and query of neuropeptides. Meanwhile, two algorithms, prescreening precursors prior to de novo sequencing (PRESnovo) and post-treatment to select potential neuropeptide candidates (HyPep) are developed to interpret low-quality MS/MS spectra and thus enhance neuropeptide discovery. With these algorithms, many new neuropeptides are discovered from mass spectral data. In proteomics, accurate peptide and protein identification and quantification is a key step to understand the role of proteins in biological processes. However, traditional database search strategy lacks sensitivity and accuracy for peptide identification. To address this issue, a bioinformatic tool (RT-SVR + q) is developed in which retention time is employed to improve peptide identification while q value metric is used to replace false discovery rate (FDR) for identification evaluation. Finally, spectral counting, a label-free protein quantification technique, is optimized to demonstrate high performance for large-scale quantitative proteomic analysis. Collectively, this body of work develops and applies bioinformatics tools in MS-based neuropeptidomics and proteomics, accelerating proteomic data analysis while enabling extraction of biological insights from complex mass spectral data via computational techniques.