Critical Transition-based Correlation Analysis for Metabolomics

摘要 代谢组提供了一个独特的窗口，以监测系统的分子状态，作为一个内在的结果， 和外部事件。采用超高效液相色谱（UPLC）高分辨率质量 高分辨质谱（HRMS），代谢组学的黄金标准技术，可以测量数百个 在数千至数百万个细胞中的代谢产物，以增强痕量化合物的信号。最近 HRMS技术的进步已经将这些测量扩展到对罕见或 珍贵的样品，取平均值不可行或妨碍结果解释。尽管有 高灵敏度HRMS，代谢组学的瓶颈是缺乏能够检测痕量水平的软件工具 信号在所产生的复杂代谢组学数据。拟议的工作填补了这一技术空白， 开发一个软件套件，调查HRMS数据集的痕迹级信号（具体目标1），并帮助找到 代谢物方差之间的相关性（具体目标2）。该方法源于手动数据处理 已建立并验证的高灵敏度分析和关键转换方案 物理学中的模型，有效地指示网络中的过渡点。该软件使用HRMS进行验证 已经获得的用于区分南非人早期发育胚胎中的细胞的数据集， 爪蟾（Xenopus laevis），细胞和发育研究以及功能实验的有力模型 测试特定代谢物的发育意义。这项工作包括旨在确保 软件与来自广泛类型质谱仪器的HRMS数据的兼容性， 不同类型的代谢组学研究，包括神经科学和药物代谢和数据储存在 MetabolomicsDatabase，一个公共的代谢组学数据库。最终产品是代谢组学软件 这适用于广泛类型的代谢组学研究。除了提供新的软件， 在这项工作中获得的代谢组学变化提供了新的信息，细胞分化的基础 发育中的胚胎。拟议工作的结果与RFA-RM-15-021的目标相匹配， 代谢组学数据分析（R 03）。所提出的软件是可扩展的HRMS数据从不同的仪器 供应商，帮助识别痕量水平的代谢物信号，并促进代谢物的分析， 系统中的代谢物相关性，这反过来又有助于设计假设驱动的研究，以帮助 更好地了解健康。

Abstract The metabolome provides a unique window to monitor a system's molecular state as a result of both intrinsic and extrinsic events. Using ultrahigh-performance liquid chromatography (UPLC) high-resolution mass spectrometry (HRMS), the gold standard technology of metabolomics, it is possible to measure hundreds of metabolites in thousands-to-millions of cells to enhance the signal for trace-level compounds. Recent advances in HRMS technology have extended these measurements to also trace-level detection for rare or precious samples, where averaging is not feasible or hinders results interpretation. Despite the availability of high-sensitivity HRMS, a bottleneck in metabolomics is a lack of software tools capable of detecting trace-level signals in the resulting complex metabolomics data. The proposed work fills this technological gap by developing a software suite that surveys HRMS data sets for trace-level signals (Specific Aim 1) and helps find correlations between metabolite variances (Specific Aim 2). The approach stems from manual data processing protocols that have been established and validated for high-sensitivity analyses as well as critical transition models in physics that efficiently indicate transition points in a network. The software is validated using HRMS data sets that have been acquired for differentiating cells in the early developing embryo of the South African clawed frog (Xenopus laevis), a powerful model in cell and developmental studies, and functional experiments that test the developmental significance of select metabolites. The work includes tests designed to ensure the compatibility of the software to HRMS data from broad types of mass spectrometry instrumentation and different types of metabolomics studies, including neuroscience and drug metabolism and data deposited in MetabolomicsWorkbench, a public metabolomics data repository. The final product is metabolomics software that is applicable to broad types of metabolomics investigations. Besides providing new software, the data that are obtained during this work provide new information on metabolomic changes underlying cell differentiation in the developing embryo. The outcomes of the proposed work are matched with the goals of RFA-RM-15-021, “Metabolomics Data Analysis (R03).” The proposed software is scalable to HRMS data from diverse instrument vendors, aids the identification of trace-level metabolite signals, and facilitates the analysis of metabolite- metabolite correlations in the system, which in turn facilitates the design of hypothesis-driven studies to help better understand health.