A 2D segmentation method for jointly characterizing epigenetic dynamics in multiple cell lines

项目摘要： 分子生物学中的一个基本问题是了解蛋白质和DNA如何相互作用以调节基因 表达和影响表型。随着先进的测序技术，大量的基因， 表观遗传学和基因组数据集已经迅速生成。利用数百个全基因组数据 许多样本的集合为我们提供了一个前所未有的机会来研究 调控标记及其对基因表达的影响。通过比较样本的全基因组特征， 在特定细胞类型中起作用的关键调节子可以用相当大的能力和分辨率来鉴定。新 可以推导并检验细胞分化过程中基因调控机制的假说， 然后将阐明以前在疾病易感性遗传学上难以解决的问题。 虽然已经进行了许多计算努力来研究表观遗传动力学和精确定位 它们的位置，一直缺乏统一和强大的框架来分析多个基因组联合在一个 这种方式解释了表观遗传事件的位置和细胞类型特异性。我们最近推出了一个新的 贝叶斯方法称为IDEAS（综合和判别表观基因组注释系统）， 解决了这一需要，并使用独立的实验数据，我们已经证明了其上级性能 现有的最先进的算法。 在这个项目中，我们的目标是大大扩展IDEAS方法的范围和适用性，并 开发一个强大的软件工具供公众使用。特别是，我们建议1）片段基因组缺失 跟踪无数据插补并整合研究之间的结果; 2）建立协变量效应模型并检测 表观基因组关联; 3）推断细粒度的局部细胞类型关系;以及4）整合染色质 构象数据，以改善分割。 在与哈迪森博士（co-I）的合作中，我们将进一步评估我们预测的一个子集的准确性 实验性的该项目的成功将有利于方法开发，产生新的资源， 重要的是，提高我们在大规模数据集成方面的能力， （epi）基因调控和复杂疾病的遗传学。

Project Summary: An essential problem in molecular biology is to understand how proteins and DNA interact to regulate gene expression and influence phenotypes. With advanced sequencing technologies, massive amount of genetic, epigenetic, and genomic data sets have been quickly generated. Exploiting the hundreds of genome-wide data sets across many samples provides us with an unprecedented opportunity to study the interplays among regulatory marks and their impacts on gene expression. By comparing genome-wide features across samples, key regulators functioning in specific cell types can be identified with substantial power and resolution. New hypotheses for the mechanisms of gene regulation during cell differentiation can be derived and tested, which will then illuminate previously intractable issues in the genetics of disease susceptibility. While numerous computational endeavors have been conducted to study epigenetic dynamics and pinpoint their locations, there has been a lack of unified and powerful framework to analyze multiple genomes jointly in a way that accounts for both position and cell type specificity of epigenetic events. We recently introduced a new Bayesian method called IDEAS (integrative and discriminative epigenome annotation system) that satisfactorily addressed this need, and using independent experimental data we have demonstrated its superior performance over existing state-of-the-art algorithms. In this project, we aim to substantially expand the scope and applicability of the IDEAS method, and to develop a powerful software tool for public use. In particular, we propose to 1) segment genomes with missing tracks without data imputation and integrate results between studies; 2) model covariate effects and detect epigenomic association; 3) infer fine-grained local cell type relationships; and 4) integrate chromatin conformation data to improve segmentation. In collaboration with Dr. Hardison (co-I), we will further evaluate the accuracy of a subset of our predictions experimentally. The success of this project will benefit method development, generate new resources, and importantly, advance our capability in large-scale data integration towards understanding the roles of (epi)genetics in gene regulation and complex disease.