Statistical methods for gene regulatory analysis and single cell genomics

项目总结 这个项目有三个主要组成部分。 分析不同环境中匹配的可访问性、表情和3D联系数据 在这个组件中，我们将开发统计方法来建模基因调控关系 关于基因表达数据(来自RNA-seq)、染色质可获得性数据(来自 ATAC-SEQ或DNase-SEQ)和来自不同蜂窝的3D交互数据(来自Hi-C或HiChIP 上下文。重点将放在不同类型的组学分析所在的“匹配数据集”上 在相同或紧密匹配的蜂窝环境中执行。方法论应该是 适用于在大量上下文子集中缺少某些数据类型的情况。这个 方法还应该能够结合大量非上下文相关的数据。 时间进程中的匹配组学数据分析 在此组件中，我们将开发一种能够对时间进程数据进行建模的方法 利用从不同背景中学习的一般监管模式提供的先验信息。 重点将放在表情、可访问性和3D联系数据 在每个时间点产生一个研究特定的生物过程。 多种类型单细胞组学数据的联合分析 在这一部分中，我们将为单细胞组学数据的联合分析开发统计方法 关于表达、可访问性和3D接触。具体地说，我们感兴趣的是 多种类型的单细胞组学数据是从相同的异质群体中产生的 细胞。我们将开发统计方法，将人口分解为相关的亚群 并推断亚群特有的基因调控关系。我们还将开发方法来 处理3D接触数据来自批量样本的情况。

PROJECT SUMMARY This project has three major components. Analysis of matched accessibility, expression and 3D contact data in diverse contexts In this component we will develop statistical methods for modeling gene regulatory relations based on the joint analysis of gene expression data (from RNA-seq), chromatin accessibility data (from ATAC-seq or DNase-seq) and 3D interaction data (from Hi-C or HiChIP) from diverse cellular contexts. The focus will on “matched data sets” where the different types of omics assays are performed on the same or closely matched cellular contexts. The methodology should be applicable to cases when some data types are missing in a substantial subset of contexts. The methodology should also be able to incorporate a large variety of non-context dependent data. Analysis of matched omics data in time courses In this component we will develop an approach to the modeling of time course data that is capable of utilizing prior information provided by a general regulatory model learned from diverse contexts. The emphasis will be on experiments where expression, accessibility and 3D contact data are generated at each time points to study a specific biological processes. Joint analysis of multiple types of single cell omics data In this component we will develop statistical methods for the joint analysis of single cell omics data on expression, accessibility, and 3D contact. Specifically, we are interested in the situation where multiple types of single cell omics data are generated from the same heterogeneous population of cells. We will develop statistical methods to resolve the population into relevant subpopulations and to infer subpopulation-specific gene regulatory relations. We will also develop methods to handle the case when 3D contact data is from bulk sample.