Analytic tools to examine high-resolution and genome-scale DNA methylation data

描述(由申请人提供)：DNA甲基化在调节哺乳动物发育过程中的谱系规范和潜能限制方面起着关键作用；DNA甲基化的异常模式通常在癌症中观察到。亚硫酸氢盐处理过的DNA的第二代测序使DNA甲基化能够得到更详细的检查，最近证明的基于阵列的捕获技术允许在选定的基因组区域进行超深度亚硫酸氢盐测序。这个项目开发了基于这些新的实验技术的数据来制定和测试关于DNA甲基化的具体假设所需的算法和统计方法。将设计一系列统计模型来表征细胞或样本中DNA甲基化的特征，并将为相关的计算任务设计算法，如模型拟合、概率计算和特征识别。这些方法将通过应用于新的超深度亚硫酸氢盐测序数据来验证；关于DNA甲基化调节和甲基化如何调节基因表达的具体假设将同时进行测试。将产生与发育和癌症相关的特定甲基化数据集。将开发用于识别细胞甲基化特征的克隆特征的计算方法，用于预测细胞之间的发育关系，以及用于解析关于肿瘤样本的复杂性和组织学的信息。这些方法的高效和健壮的实现将被开发并发布给公众使用。拟议的研究将提供更多的分析能力，以补充研究DNA甲基化的新兴实验技术。这将使研究人员，特别是那些研究人类发育和癌症的研究人员，能够提出和回答关于DNA甲基化功能的更准确的问题。此外，这项技术将有助于确定与癌症相关的临床结果的最有效的甲基化标记。

DESCRIPTION (provided by applicant): DNA methylation plays a critical role in regulating lineage specification and restriction of potency during mammalian development; aberrant patterns of DNA methylation are generally observed in cancers. Second-generation sequencing of bisulfite treated DNA is enabling DNA methylation to be examined in greater detail, and recently demonstrated array-based capture technique allow ultra-deep bisulfite sequencing in selected genomic regions. This project develops algorithmic and statistical methods required to formulate and test specific hypotheses about DNA methylation based on data from these novel experimental technologies. A family of statistical models will be designed to characterize features of DNA methylation in a cell or sample, and algorithms will be designed for associated computational tasks of model fitting, probability calculations and feature identification. The methods will be validated through application to novel, ultra-deep bisulfite sequencing data; specific hypotheses about the regulation of DNA methylation and how methylation regulates gene expression will be tested simultaneously. Specific methylation datasets related to development and cancer will be produced. Computational methods will be developed for identifying clonal features of methylation profiles in cells, for predicting developmental relationships between cells, and for resolving information about the complexity and histology of tumor samples. Efficient and robust implementations of the methods will be developed and released for public use. The proposed research will provide increased analytical capability to complement emerging experimental technology for investigating DNA methylation. This will enable researchers, particularly those studying human development and cancers, to ask and answer more precise questions about the functions of DNA methylation. Moreover, this technology will assist in identifying the most effective methylation-based markers for clinical outcomes associated with cancers.