Integrative analyses of reference epigenomic maps and applications

描述（由申请人提供）：最近在大规模数据生成项目的努力中，试图生成多种参考组织和癌症组织的基因组和表观基因组图谱（表观基因组路线图，ENCODE， IHEC， TCGA）。这些正交的“组学”图谱可用于了解正常组织中的组织特异性调节，以及识别导致疾病的失调区域。然而，不同的分析方法产生的数据通常是异构的，每个特定的平台都有不同的技术偏差。此外，当前的规范化和人工校正方法是特定于平台和数据类型的，并且可能需要应用的训练集和测试集，从而导致数据过度拟合，并且如果稍后引入额外的样本，则会导致重新规范化或分析修改。在这项研究中，我们将开发和展示一个全面和协调的框架，用于跨组织类型和分析平台的数据标准化。我们将开发用于单个分析样本统一预处理的算法方法，从而实现跨平台、实验和技术的无缝集成和解释。我们将使用我们的标准化数据来识别跨多种组织/细胞类型和跨表观基因组分析数据平台的协调和不协调的表观遗传共调控区域。我们特别关注上皮细胞类型。最后，我们将开发跨多种组织或细胞类型的表观遗传药物的多“组学”药物功效生物标志物，随后在来自多个公共数据资源的患者样本上进行演示。

DESCRIPTION (provided by applicant): Recent efforts in large-scale data generation projects have attempted to produce genomic and epigenomic maps of multiple reference and cancer tissues (Epigenomic Roadmap, ENCODE, IHEC, TCGA). These orthogonal '-omic' profiles can be used to understand tissue-specific regulation in normal tissues, as well as identify regions of dysregulation that lead to disease. However, the data produced by different profiling approaches are typically heterogeneous and each specific platform suffers from different technical biases. Furthermore, current normalization and artifact correction methods are platform and data type specific, and may require both the training and test set for application, leading to data over-fitting, and renormalization or analysis modification if additional samples are later introduced. I this research, we will develop and demonstrate a comprehensive and coordinated framework for data standardization across tissue types and profiling platforms. We will develop algorithmic approaches for the uniform preprocessing of individual profiling samples that allow for seamless integration and interpretation across platforms, experiments, and techniques. We will use our standardized data to identify regions of coordinate and discordant epigenetic co-regulation across multiple tissue/cell types and across epigenomic profiling data platforms. Our particular focus will be across epithelial cell types. Finally, we will develop multi-'omic' drug efficacy biomarkers for epigenetic drugs across multiple tissue or cell types, followed by demonstration on patient samples from multiple public data resources.