Chemometric Strategies for the Analysis of Multivariate Chemical and Biological Measurements

分析测量科学的发展使我们对生物系统和医学的理解取得了前所未有的进步。在复杂系统中以不断降低的浓度水平测量不断增加的生物分子的能力为了解生物有机体的功能提供了新的见解，并且这些能力将继续扩大。然而，在所谓的“组学”科学中收集和解释相关数据仍然是高通量生物学方法的一个挑战。这项研究通过开发新的方法来可视化和解释海量数据，将测量转化为信息来解决这些挑战。此外，还提出了新的策略，以提高蛋白质组学方法的吞吐量和实用性。这项工作的核心是利用申请人在化学计量学和生物信息学领域的专门知识，这两个领域分别涉及应用统计、数学和计算机方法从化学和生物测量中提取信息。现代分析系统通常提供对数千种化学实体(遗传物质、蛋白质、代谢物等)的测量。在单个生物样本中，称为多变量或高维数据。即便如此，寻求的关键信息(例如疾病生物标志物的识别)可能无法获得，原因包括：(1)数据中没有关键分子(S)，(2)无法区分重要信号(S)和无意义的信号，以及(3)由于分析时间长，样本数量在统计上不足。本研究将通过加强组学研究中使用的测量和数据分析方法来解决这些问题，从而提取更可靠的信息。最终，这有望对生物学和医学的研究产生重要影响。

Developments in analytical measurement science have led to unprecedented advances in our understanding of biological systems and medicine. The ability to measure increasing numbers of biological molecules at decreasing concentration levels in complex systems provides new insights into the function of biological organisms, and these capabilities will continue to expand. However, the collection and interpretation of relevant data in the so-called "omics" sciences remains a challenge for high-throughput biological methods. This research addresses those challenges through the development of novel approaches to visualize and interpret the massive amounts of data, transforming measurements into information. Additionally, new strategies are proposed to increase the throughput and utility of methods employed in proteomics. The core of this work draws on the applicant's expertise in the fields of chemometrics and bioinformatics, which relate to the application of statistical, mathematical and computer-based methods for the extraction of information from chemical and biological measurements, respectively. Modern analytical systems typically provide measurements on thousands of chemical entities (genetic material, proteins, metabolites, etc.) in a single biological sample, referred to as multivariate or high-dimensional data. Even so, the critical information sought (e.g. identification of a disease biomarker) may not be obtained for several reasons that include: (1) the absence of the critical molecule(s) in the data, (2) the inability to distinguish the important signal(s) from the meaningless ones, and (3) a statistically insufficient number of samples due to long analysis times. This research will address these issues by enhancement of the measurement and data analysis methods used omics research, leading to the extraction of more reliable information. Ultimately, this is expected to have important implications for the study of biology and medicine.