An Integrative Approach to Construct a Regulatory Network Effected by TDZs

描述（由申请方提供）：肥胖是代谢紊乱的主要风险因素。肥胖通常导致功能失调的白色脂肪组织（WAT）的积累，这进一步导致代谢失调，脂肪酸循环升高和促炎脂肪因子分泌增加。人类脂肪燃烧棕色脂肪组织（BAT）的发现使BAT成为治疗肥胖和代谢性疾病的新方法。噻唑烷二酮（TZD）具有将WAT转化为“类棕色”状态的功能。此外，TZDs已被用作糖尿病的补救措施。但是TZDs的临床应用受到限制， 安全性问题，如潜在的心血管风险。了解机制将确定有效的，但风险较低的药物代谢紊乱的目标。TZD通过激活PPAR发挥作用？（过氧化物酶体增殖物激活受体？）。然而，我们对PPAR靶点的了解？和其他辅因子是有限的。了解TZDs的作用，进一步研究TZDs对布朗宁的影响 效果，我们建议开发一种新的算法来预测远程启动子-增强子相互作用，并构建一个转录网络。为了预测长程相互作用，我们将采用一种机器学习算法，该算法使用从全局连续测序（GROseq）数据中获得的增强子RNA（eRNA）水平和基因转录水平。GROseq是预测长程相互作用的有用数据集，因为eRNA水平与基因转录水平高度相关。将获得的相互作用应用于包括PPAR？在内的TF的已知结合位点，GR，C/EBP，SMRT和RXR，我们将构建一个全面的TF-基因网络。预测的相互作用提供了一个有用的环境，研究TZDs的基因调控。我们将研究距离，相对位置，以及多个TF结合位点的组合如何影响基因表达。我们还将通过在网络中包含BAT特异性TF结合数据来研究TZD的布朗宁效应。从TF-基因网络中分析BAT特异性结合信息的转录规律，并结合其他TF进行分析。总体而言，这些研究采用创新和创造性的方法整合各种类型的数据来研究TZDs的基因调控。该算法通过对复杂基因组数据进行再处理，构建了一个完整的调控网络，为分析布朗宁效应的调控规律提供了独特的视角，这将极大地加深我们对TZDs基因调控的理解，并为糖尿病的治疗寻找潜在的靶点。

DESCRIPTION (provided by applicant): Obesity is a major risk factor for metabolic disorders. Obesit typically leads to accumulation of dysfunctional white adipose tissue (WAT), which further causes metabolic dysregulation with elevated circulation of fatty acids and increased secretion of proinflammatory adipokines. The discovery of fat burning brown adipose tissue (BAT) in humans has raised the exciting possibility of BAT may be targeted as a novel method to treat obesity and metablic diseases. Thiazolidinediones (TZDs) have a function to convert WAT into a "brownlike" state. Beside, TZDs have been used as a remedy for diabetes. But the clinical use of TZDs has been limited because of the safety concerns such as potential cardiovascular risks. Understanding the mechanism will identify efficacious but lower risk drug targets for the metabolic disorders. TZDs act by activatin PPAR? (peroxisome proliferator-activated receptor ?). However, our understanding about the targets f PPAR? and other cofactors is limited. To understand the role of TZDs and further study the browning effect, we propose to develop a novel algorithm to predict long-range promoter-enhancer interaction and construct a transcriptional network. To predict the long-range interactions, we will employ a machine learning algorithm that uses the enhancer RNA (eRNA) levels and gene transcription levels obtained from global run-on sequencing (GROseq) data. GROseq is a useful dataset to predict long-range interactions, as the eRNA levels highly correlate with the gene transcription level. Applyingthe obtained interactions to the known binding sites of TFs including PPAR?, GR, C/EBP, SMRT, and RXR, we will construct a comprehensive TF-gene network. The predicted interaction provides a useful environment to study gene regulation of TZDs. We will study how the distance, relative position, an the combination of multiple TF binding sites affect gene expression. We will also investigate the browning effects by TZDs by including BAT-specific TF binding data in the network. The transcriptioal rule of the BAT-specific binding information, in combination with other TFs, will be analyzed from he TF-gene network. As a whole, these studies use an innovative and creative approach to integrate various types of data to study gene regulation of TZDs. By reprocessing complex genomic datasets ino a comprehensive regulatory network, the proposed algorithm provides a unique view in analyzing the regulatory rules of the browning effect, which will greatly enhance our understanding about the gen regulation of TZDs and identify potential therapeutic targets for diabetes.