Quantitative Methods for Genome-wide Analysis of Macrophage Activation by ESCs

描述(申请人提供)：该提案旨在推进基因组研究的新统计方法和发现胚胎干细胞(ESCs)移植后畸胎瘤发生和发展的复杂分子机制。其生物学目的是研究胚胎干细胞诱导的巨噬细胞激活和畸胎瘤发生的机制和生物学途径，揭示人类MIF基因多态性与血清MIF浓度的关系，以选择安全的细胞移植患者。为了应对这些挑战，高通量基因组工具和新的统计方法将结合起来，以加强我们对巨噬细胞在畸胎瘤发生发展中的作用的理解。这项研究不仅将专注于识别巨噬细胞上的单个基因变化，而且还将分析潜在的生物途径。此外，动物模型中的时程分析将导致从畸胎瘤发展的早期到晚期的不同过渡时期的基因识别。这些生物学研究在统计学上提出了许多新的挑战。特别是，提出了新的统计技术来回答以下重要问题：如何提高现有检验对具有稀疏性的大规模多重检验问题的能力；如何检测基因是否随时间过程表达；如何使用定量工具发现分子机制；如何在一般相关性下控制误发现率；信号在任意相关性下的最佳检测边界；如何估计超高维回归中的残差方差。这些主题在压缩传感和计算机安全方面也有重要的应用。此外，还提出了几种稳健有效的超高维稀疏回归、分类和遗传联网方法。新提出的方法将被应用于全基因组关联研究，以确定与MIF表达相关的单核苷酸多态，以便选择安全的干细胞移植患者。该建议的一个重要特征是综合了PI在统计学和细胞生物学方面的广泛知识，以更好地了解复杂的生物学问题。

DESCRIPTION (provided by applicant):The proposal intends to advance both novel statistical methods for genomic research and discoveringcomplex molecular mechanisms on the teratoma initiation and progression after transplantation ofembryonic stem cells (ESCs). The biological aims are to investigate the mechanisms and biologic pathwaysof ESC-induced macrophage activation and teratoma development; and to unveil how the human MIFpolymorphisms associated with the serum MIF concentration in order to select patients for safe celltransplantation. To meet these challenges, high-throughput genomic tools and novel statistical methods willbe combined to enhance our understanding on the macrophage function on initiation and progression ofteratomas. The research will not only focus on identifying individual gene change on macrophages, but alsoanalyze underlying biologic pathways. In addition, time-course analysis in animal models will lead to identifygenes in different transition periods from early- to late-stage of teratoma development. These biologicalstudies prompt many new challenges in statistics. In particular, novel statistical techniques are proposed toanswer the following important questions: how to improve the power of existing tests for large-scale multipletesting problems with sparsity; how to detect whether genes have expressed over time course; how to usequantitative tools for discoveries of molecular mechanisms; how to control false discovery rate undergeneral dependence; what optimal detecting boundaries of signals are under arbitrary dependence; how toestimate residual variance in ultrahigh-dimensional regression. These topics have also importantimplications on compress sensing and computer security. In addition, several robust and effective methodsare proposed for ultrahigh-dimensional sparse regression, classification, and genetic networking. Thesenewly proposed methods will be applied to genome-wide association studies to identify single-nucleotidepolymorphisms associated with MIF expression in order to select patients for safe stem cell transplantation.An important feature of the proposal is the synthesis of PIs' extensive knowledge in statistics and cellbiology to gain better understanding of complicated biological problems.