SEI: Unraveling the Structure and Kinetics of Biochemical Pathways from Time-Series Analysis

随着基因组科学中新的高通量实验技术的快速发展，基因、蛋白质和代谢物浓度的基因表达时间序列正在变得可用。这种时间序列隐含着关于潜在遗传或生化网络机制的连通性和调控结构的有价值的信息。提取这种信息是一项具有挑战性的任务，因为它需要发展新的非线性估计的数学和计算方法，其中包括迭代搜索算法。用高质量的初始猜测来启动这些算法可以极大地加速搜索过程。即使当一个生物体的完整基因组序列可用时，也只有一小部分基因成分的功能和相互作用是清楚的。目前，未知蛋白质的功能通常是根据序列相似性、共同结构基序、基因顺序、基因融合事件或基因表达的相似性来推断的。该提案的目标是开发一种基于基因在网络中的作用进行功能预测的新方法。这种方法使我们能够独立于结构或序列上的同源性对蛋白质进行功能预测，并提供了一种使用高通量技术已发表的生物学数据来表征尚未研究的蛋白质的方法。该方法将适用于任何生物，包括人类，在这些生物中，只有3%到5%的基因功能已知。随着我们更好地了解基因和蛋白质在网络环境中的功能，我们就可以更好地预测和控制它们对内部和外部扰动的反应。在可预见的未来，建模预测的类型很可能是制药业和生物医学科学决策过程中的许多输入之一。这项研究将为本科生、研究生和博士后研究人员提供跨学科(生物学、数学和计算；实验和理论)培训，并帮助培养出既熟悉生物学、数学又熟悉计算的一代科学家。

Time series of gene expression of gene, protein and metabolite concentrations are becoming available as the result of the rapid development of novel, high-throughput experimental techniques in genomics sciences. Such time series implicitly contain valuable information about the connectivity and regulatory structure of the underlying genetic or biochemical network mechanism. The extraction of this information is a challenging task because it requires the development of new mathematical and computational methods of nonlinear estimation that involve iterative search algorithms. Priming these algorithms with high-quality initial guesses can great accelerate the search process.Even when full genomic sequences for an organism are available, the functions andinteractions of only a small number of gene components are clear. Presently, the functions of uncharacterized proteins have usually been inferred on the basis of sequence similarities, common structural motifs, gene order, gene fusion events, or similarities in gene expression. The proposal goal is to develop a new method for functional predictions based on the role of the gene in networks. This method allow us to perform functional predictions for proteins independent of homologies in structure or sequence, and provide a way to characterize proteins that have not yet been studied using published biological data from high-throughput technologies.The methodology will be applicable to any organism, including humans, whereonly three to five percent of gene function is known. As we better understand the functions of genes and proteins in a network context, we can better predict and control their responses to internal and external perturbations. For the foreseeable future, the type of modeling predictions will likely be one of the many inputs into the decision making process in the pharmaceutical industry, and biomedical sciences. The research will provide interdisciplinary (biological, mathematical and computational; experimental and theoretical) training to undergraduate and graduate students and postdoctoral researchers, and help produce a generation of scientists comfortable both with biology, mathematics and computation.