Inferring Mammalian Transcriptional Regulatory Networks from Epigenomics

描述（由申请人提供）：破译哺乳动物系统中控制生物过程的转录调控网络（TRN）对于我们理解正常生理学和疾病病因学的基本机制至关重要。这是一项艰巨的任务，因为TRN中有太多的链路是未知的。ChIP-chip和ChIP-seq技术的出现使得能够对已知是生物过程中的关键调节因子的许多转录因子（TF）的全基因组结合位点进行定位。然而，这两种技术仅限于具有ChIP质量抗体的已知调节剂。我们发现TF结合通常与动态组蛋白标记签名相关，并且可以从全基因组蛋白标记动态计算预测。因此，我们假设，随着时间进程的核小体分辨率ChIP-seq的一些信息组蛋白标记和基因表达的RNA-seq数据，和有效的计算建模，我们可以推断哺乳动物生物过程中的TRN。具体来说，我们建议开发有效的计算算法，以实现目标1：第一，预测TF结合从核小体分辨率组蛋白标记动力学;第二，从TF结合，组蛋白标记和基因表达谱识别靶基因;第三，推断在一个时间过程中的TRN。我们还建议将上述算法应用于目标2中的两个生物系统。一个是小鼠成肌细胞系C2 C12分化成骨、脂肪或肌肉，另一个是人顶浆分泌乳腺癌细胞系MDA-MB-453在维生素D处理下可逆重编程为上皮细胞。通过时程核小体分辨组蛋白标记ChIP-seq和RNA-seq分析，我们将计算推断和实验验证这两个系统中的TRN。

DESCRIPTION (provided by applicant): Deciphering the transcriptional regulatory network (TRN) governing a biological process in mammalian systems is essential to our understanding of basic mechanisms underlying normal physiology as well as disease etiology. It is a daunting task because too many links in the TRN are unknown. The emergence of ChIP-chip and ChIP-seq technologies has enabled the mapping of the genome-wide binding sites of many transcription factors (TFs) known to be key regulators in a biological process. However, these two technologies are limited to the known regulators with ChIP-quality antibodies. We found that TF binding is often associated with a dynamic histone mark signature and can be computationally predicted from the genome-wide histone mark dynamics. Therefore, we hypothesize that with time-course nucleosome-resolution ChIP-seq of a few informative histone marks and RNA-seq data of gene expression, and effective computational modeling, we could infer the TRNs in mammalian biological processes. Specifically, we propose to develop effective computational algorithms to achieve Aim1: first, predict TF binding from nucleosome-resolution histone mark dynamics; second, identify target genes from TF binding, histone marks and gene expression profiles; and third, infers the TRN over a time course. We also propose to apply the above algorithms in two biological systems in Aim 2. One is the mouse myoblast cell line C2C12 differentiation into bone, fat, or muscle, and the other is the human apocrine breast cancer cell line MDA-MB-453 reversible reprogramming to epithelial cells with vitamin D treatment. Through time-course nucleosome-resolution histone mark ChIP-seq and RNA-seq profiling, we will computationally infer and experimentally validate the TRNs in these two systems.