Statistical Methods For Annotating Repetitive Genomic Regions Through ENCODE-deri

描述（由申请人提供）：ENCODE项目通过应用高通量下一代测序（NGS）产生了丰富的高质量基因组数据集，以创建人类和模式生物基因组中的功能元件目录。尽管ENCODE采用的NGS技术能够以公正的方式对基因组进行分析，但ENCODE项目的数据分析工作迄今为止主要集中在基因组的可映射区域，因此没有充分利用这些数据来发挥其全部优势。全面理解ENCODE项目数据的一个主要瓶颈是缺乏能够识别重复区域中的功能元素的统计和计算方法。我们将利用我们在ChIP-seq和RNA-seq分析方面的专业知识，在四个特定目标中解决这一关键障碍。在目标1中，我们将开发概率模型和配套软件，用于利用从多种类型的*-seq数据集（ChIP-, DNase-， MeDIP-和FAIRE-seq）中映射到基因组上多个位置的读取（multi-reads）。这将使在重复区域的调控元件编目成为可能。在目标2中，我们将通过同时利用多个相关的*- seq数据集，从我们的概率模型中提高重复区域发现的特异性。具体来说,我们

DESCRIPTION (provided by applicant): The ENCODE projects have generated a wealth of high-quality genomic datasets with the applications of high- throughput next generation sequencing (NGS) to create a catalog of functional elements in the human and model organism genomes. Although the NGS technologies, embraced by ENCODE, are enabling interrogation of genomes in an unbiased manner, the data analysis efforts of the ENCODE projects have thus far focused on mappable regions of the genomes and thereby have not fully leveraged these data to their full advantage. A major bottleneck to a comprehensive understanding of data from the ENCODE projects is the lack of statistical and computational methods that can identify functional elements in repetitive regions. We will address this critical impediment in four specifi aims by building on our expertise in ChIP-seq and RNA-seq analysis. In Aim 1, we will develop probabilistic models and accompanying software for utilizing reads that map to multiple locations on the genome (multi-reads) from multiple types of *-seq datasets (ChIP-, DNase-, MeDIP-, and FAIRE-seq). This will enable cataloging of regulatory elements in repetitive regions. In Aim 2, we will improve the specificity of the discoveries in repetitive regions from ou probabilistic models by utilizing multiple related *- seq datasets simultaneously. Specifically, we will devise methods to supervise analysis of ChIP- and RNA-seq datasets by external ChIP-seq datasets. This will facilitate accurate inference for repetitive elements with near identical sequences, e.g., segmental duplications, long interspersed nuclear elements, and boost accuracy of gene and isoform quantification with RNA-seq. In Aim 3, we will focus on identifying co-occupied/enriched regions to infer cell-specific modules of regions/genes and their regulatory profiles. We will also develop a formal differential co-enrichment framework to study cell-specific wiring and interactions of regulatory factors. This will elucidate how interactions among regulatory factors vary across cells/tissues/conditions. Aim 4, we will apply our methods from Aims 1-3 to relevant ENCODE data to understand GATA factor functions in hematopoiesis and vascular biology. The GATA system in human and mouse will serve as a training and validation platform for our methods. Statistical and computational resources generated from the project, which will be disseminated as modular and robust software, will help to enhance and maximize the impact of ENCODE-derived data on the biomedical research community.