白血病多层次转录调控组学数据的信息整合与可视化挖掘

The initiation and progression of leukemia is not only driven by the genetic changes such as gene mutations and chromosomal translocations, but is also under the control of epigenetic regulations that involve DNA methylation and chemical markers on core histone tails. The advances of the high-throughput technologies revolutionize the studies on leukemogenesis, allowing for various biological data produced in parallel and on a genome scale (omics). However, it still remains a great challenge to make sense of these multi-layer omics data. Such challenge is largely due to the noisy nature of data and the difficulty in clearly defining causative relationships between histone modifications and regulatory outcome. In this regard, the self-organizing map (SOM) is particularly useful, as we have shown previously, for topology-preserving characterization of multi-dimensional omics data. As proposed here, we aim to establish a SOM-centric analytical pipeline and web server for maximizing its beneficial potentials in visualizing, clustering and integrating omics data of various types. Using the proposed pipeline together with databases of ontologies and regulatory elements, we also aim to decode the inherent relationships among multi-layer regulatory omics data in leukemia, including but not limited to key transcription factor binding sites (i.e., cistrome), histone epigenome and transcriptome. Last but not least, we expect to identify novel signatures integrating histone modifications and expression level, and finally evaluate clinical value of these integrative signatures via correlating with leukemia patients, both in silico and from bench.

白血病的发生发展不仅涉及基因突变、染色体易位形成的融合蛋白等遗传学改变，而且还涉及DNA甲基化、组蛋白修饰等表观遗传学机制。高通量组学技术为研究白血病提供技术保障与多层次转录调控组学数据，但同时也给相应的信息整合与挖掘带来了巨大的挑战，如数据本身假阳性与假阴性、转录调控与表观修饰关系难以界定等。本项目拟在前期自主开发并在解析多个生物学模型的机制方面展现出多种优势的基础上，利用自组织图（Self-organizing map）在规律性模式识别、去除干扰信号与多维可视化展示的特色，结合各种信息注解、调控元件等数据库，建立一套较完备的多层次组学信息整合与可视化挖掘框架，重点解析项目组已积累的白血病关键转录因子结合位点、各种组蛋白表观修饰、转录表达水平之间内在的关系。同时，应用该系统进一步挖掘若干组蛋白修饰与基因表达的联合标识，分析并实验验证其在白血病病人样本中的潜在临床价值。

项目背景: 下一代测序技术产生白血病多层次转录调控组学数据。组学数据视为数据云点，将聚集于某一对称中心，反映大部分基因没有发生变化这一特点。.主要内容： 受到天然形成特殊结构（例如蜂巢）的启发，本课题组创造性地构建“超正六边形图”（supra-hexagonal map）这一高度对称的拓扑结构，并进而衍生出一系列2D拓扑结构，包括指环状（ring）、钻石（diamond）、三叶草（trefoil）、蝶状（butterfly）、沙漏（hourglass）、梯状（ladder）、桥形（bridge）以及三角形（triangle）。在这些拓扑结构上实现自组织算法（self-organising algorithm; SOM），开发基于SOM多层次生物组学信息整合与可视化挖掘框架。.重要结果：收集并分析各种白血病细胞模型相关组学数据，尤其是关键转录因子在全基因组水平ChIP-seq结合位点数据、DNaseI超敏位点（DNaseI hypersensitivity site）数据以及其它调控组学数据。证明研发的挖掘框架适用于大数据整合分析任务，既适用于不同类型数据也适用于不同数据对象。由于可视化的优势，多层次组学数据之间的内在关系可以直观展示，揭示转录因子结合分布同染色质活性状态关系、转录因子在不同细胞系差异结合分布的结构基础以及细胞间转分化途径等。.关键数据：软件包supraHex发布于公共平台（http://r-forge.r-project.org/projects/suprahex)，在2016年度共下载超过1500次（独立IP地址；数据来自于Bioconductor网站）。科研成果论文“A collection of aesthetic maps for the self-organised representation and comparison of big data in regulatory genomics (are insightful)"目前正在同行评审中。.科学意义：贡献了一种组学大数据分析的研究方法。发现了诸多潜在的关键转录因子，为后续湿实验验证提供丰富素材，也为细胞状态分类以及细胞转分化等预测应用提供一种可能。