Model-Based Methods for Analyzing ChIP Sequencing Data

描述(申请人提供)：蛋白质-DNA相互作用是基因调控靶基因表达的基本机制。破译这一机制是具有挑战性的，因为在基因组规模上表征蛋白质结合的DNA是困难的。最近超高通量测序技术的到来使这一领域发生了革命性的变化，使目标DNA的定量测序能够以快速和经济有效的方式进行。ChIP-Seq将染色质免疫沉淀(ChIP)与下一代测序相结合，提供数百万个短读序列，代表特定转录因子和其他染色质相关蛋白结合的DNA标签。芯片序列数据的快速积累带来了令人望而生畏的分析挑战。在这里，我们提出了一种基于隐马尔可夫模型(HMM)的算法来检测显着富含CHIP-SEQ的基因组区域。我们的方法将解决诸如测序偏差和读取比对不确定性等复杂问题。我们还提出了一种多级分层隐马尔可夫模型，它将允许集成来自芯片序列和芯片-芯片的数据。接下来，我们将利用芯片序列数据建立基于模型的从头基序发现策略。我们相信，对CHIP-SEQ确定的所有序列的有效挖掘可以使我们准确地表征蛋白质-DNA相互作用位点。我们对前列腺癌的长期生物医学研究感兴趣。我们将应用CHIP-SEQ和本项目中开发的数据分析工具来研究前列腺癌转录(Dys-)调控。我们相信，在一个连贯的概率框架下进行有效的数据整合，最终将导致对前列腺癌进展中介导转录调控的机制的深入理解。 公共卫生相关性：转录调控在癌症进展中起着重要作用。这里提出的统计和计算策略的发展将帮助我们深入了解前列腺癌进展过程中介导转录调控的机制。

DESCRIPTION (provided by applicant): Protein-DNA interaction constitutes a basic mechanism for genetic regulation of target gene expression. Deciphering this mechanism is challenging due to the difficulty in characterizing protein-bound DNA on a genomic scale. The recent arrival of ultra-high throughput sequencing technologies has revolutionized this field by allowing quantitative sequencing analysis of target DNAs in a rapid and cost-effective way. ChIP-Seq, which couples chromatin immunoprecipitation (ChIP) with next-generation sequencing, provides millions of short-read sequences, representing tags of DNAs bound by specific transcription factors and other chromatin-associated proteins. The rapid accumulation of ChIP-Seq data has created a daunting analysis challenge. Here we propose a hidden Markov model (HMM)-based algorithm to detect genomic regions that are significantly enriched by ChIP-Seq. Our method will address complications such as sequencing bias and read alignment uncertainty. We also propose a multi-level hierarchical HMM that will allow integration of data from both ChIP-Seq and ChIP- chip. Next, we will build model-based de novo motif finding strategies that utilizing ChIP-Seq data. We believe efficient mining of all sequences identified by ChIP-Seq allows us to precisely characterize the protein-DNA interaction sites. Our long term biomedical research interest is in prostate cancer. We will apply ChIP-Seq and the data analysis tools developed in this project to investigate prostate cancer transcription (dys-) regulation. We believe effective data integration under a coherent probability framework will eventually lead to an in-depth understanding of mechanisms mediating transcription regulation in prostate cancer progression. PUBLIC HEALTH RELEVANCE: Transcription regulation plays an important role in cancer progression. The development of statistical and computational strategies proposed here will help us gain in-depth understanding of mechanisms mediating transcriptional regulation in prostate cancer progression.