Integrative analysis of genomic and epigenomic datasets in multiple cell types

描述(由申请人提供)：本申请涉及广泛的挑战领域(08)基因组学，以及特定的挑战主题08-OD-101，表观基因组分析的计算方法。虽然人类基因组的初级DNA序列最终负责每个细胞的编码和功能，但近年来已经描述了大量的染色质和DNA修饰，可以调节对这个初级序列的解释。这些表观遗传修饰导致不同类型的人类细胞功能的多样性，并在发育过程中细胞身份的建立和维持以及在健康和疾病中发挥关键作用。人类ENCODE项目、NIH表观基因组路线图和其他几项大规模实验工作目前正在进行中，以绘制多种人类细胞类型和疾病状态的数十个组蛋白和DNA修饰，产生各种丰富的表观基因组数据集。这就迫切需要开发严格的计算方法，对表观基因组数据集进行系统的综合分析，并了解它们与其他基因组数据集的关系，包括基因表达、疾病关联和表型特征。在这个提案中，我们将开发和应用基于多变量隐马尔可夫模型的图形概率模型来描述染色质的修饰。我们将使用这些模型来发现一组潜在的染色质状态，基于整个基因组上反复出现的表观遗传标记组合(目标1)。我们将根据这些状态相对于现有功能元件的丰富程度和位置偏差，以及大规模的基因表达和疾病关联数据集(目标2)来验证和功能描述这些状态。最后，我们将扩展这些方法来研究健康和疾病细胞类型中染色质状态的动态，并研究这些与观察到的细胞类型之间的功能差异之间的相关性(目标3)。我们将与参与数据生产的科学家密切合作，促进他们之间的沟通和数据集成，并与已经建立的数据分析和协调中心密切合作，以促进在ENCODE和Eigenome路线图联盟以及更大的社区之间共享方法和结果。总体而言，拟议的大规模基因组和表观基因组数据集的综合分析将提供当前和计划中的表观基因组数据集的统一视图，从而系统地理解健康和疾病中的基因和基因组调控。虽然人类基因组的初级DNA序列最终负责每个细胞的编码和功能，但近年来描述了大量的染色质和DNA修饰，这些修饰可以调节这个初级序列的解释，导致不同类型的人类细胞功能的多样性。该项目将创建一个计算框架和资源，以整合大规模基因组和表观基因组数据集，了解它们在健康和疾病中的功能作用，并了解它们在不同细胞系和疾病状态中的动态。所获得的知识可以在理解健康发育过程中细胞身份的建立和维持以及这些过程的失调如何导致疾病发生方面发挥关键作用。

DESCRIPTION (provided by applicant): This application addresses broad Challenge Area (08) Genomics, and specific Challenge Topic, 08-OD-101, Computational approaches for epigenomic analysis. While the primary DNA sequence of the human genome is ultimately responsible for the encoding and functioning of each cell, a plethora of chromatin and DNA modifications have been described in recent years that can modulate the interpretation of this primary sequence. These epigenetic modifications lead to the diversity of function across different human cell types, and play key roles in the establishment and maintenance of cellular identity during development, and also in health and disease. The human ENCODE project, the NIH Epigenome Roadmap, and several other large-scale experimental efforts are currently underway to map dozens of histone and DNA modifications across multiple human cell types and disease states, generating a diversity of rich epigenomic datasets. This creates a pressing need for the development of rigorous computational methods for the systematic integrative analysis of epigenomic datasets, and for understanding their relationship to other genomic datasets, including gene expression, disease association, and phenotypic profiling. In this proposal, we will develop and apply graphical probabilistic models for describing chromatin modifications, based on multivariate hidden Markov models. We will use these models to discover the set of underlying chromatin state, based on recurrent combinations of epigenetic marks across the entire genome (Aim 1). We will validate and functionally characterize these states based on their enrichments and positional biases with respect to existing functional elements, as well as large-scale gene expression and disease association datasets (Aim 2). Lastly, we will extend these methods to study dynamics of chromatin state across both healthy and disease cell types, and study how these correlate with functional differences between the observed cell types (Aim 3). We will work closely with the scientists involved in data production and facilitate communication and data integration across them, and also with data analysis and coordination centers already established to facilitate sharing of methods and results across the ENCODE and Epigenome Roadmap consortia, and with the larger community. Overall, the proposed integrative analysis of large-scale genomic and epigenomic datasets will provide a unified view of current and planned epigenomic datasets, towards a systematic understanding of gene and genome regulation in health and disease. While the primary DNA sequence of the human genome is ultimately responsible for the encoding and functioning of each cell, a plethora of chromatin and DNA modifications have been described in recent years that can modulate the interpretation of this primary sequence, leading to the diversity of function across different human cell types. This project will create a computational framework and resource to integrate large-scale genomic and epigenomic datasets, to understand their functional role in health and disease, and to understand their dynamics across different cell lines and disease states. The knowledge gained can play key roles in understanding the establishment and maintenance of cellular identity during healthy development, and how dysregulation of these processes can lead to the onset of disease.