Computational Tools for Single Cell Analysis: Application to Retinal Degeneration

确定视网膜变性过程中改变的调节网络将为深入了解其机制提供帮助。 潜在的视网膜疾病。在单细胞水平上分析这种网络扰动将有助于我们查明 在使用批量样品的传统分析中可能错过的关键分子事件。但要 确定视网膜疾病中改变的网络仍然具有挑战性，部分原因是缺乏强大的 计算工具。首先，许多聚类方法产生不同的，有时相互矛盾的结果。第二、 单个细胞表达可用于检测以前无法识别的细胞类型，而当前的聚类 算法通常不够灵敏，不能检测新的、有时是罕见的细胞类型。第三，基因组 从大量样品和单细胞获得的相互作用可能是彼此互补的， 反映了基因调控、共表达和蛋白质-蛋白质相互作用方面的不同方面。小说 需要综合方法来最大化我们从这些基因组数据集获得的信息。解决 为了应对这些挑战，我们将开发用于单细胞数据分析的计算方法。具体来说，我们将 开发用于单细胞基因表达分析的迭代聚类方法（目的1）。我们的做法是 设计坚固而灵敏。然后，我们将开发一种方法来确定活跃的监管网络， 整合单细胞RNA-Seq数据集和来自批量样品的ATAC-Seq（目标2）。这种方法将使我们 来识别单细胞水平的调节回路。然后，我们将在视网膜中进行单细胞RNA-Seq。 退化模型（目标3），并将我们的计算方法应用于数据集。我们希望能找出 驱动程序和途径参与感光细胞变性。最后，我们将开发一个数据库， 标记基因通过分析可获得的单细胞数据集（Aim 4）。我们相信，计算 我们建议开发的算法和数据库将是研究界的宝贵资源。在- 对视网膜退行性疾病模型的深入研究将揭示导致疾病的关键分子事件， 为视网膜变性疾病提供了新的治疗靶点。

Identifying the regulatory networks altered during retinal degeneration will provide insights into the mechanisms underlying retinal disease. Analysis of such network perturbation at the single cell level will help us to pinpoint the key molecular events that could be missed in traditional analysis using the bulk samples. However, to identify the altered networks in retinal disease is still challenging, partly due to the lack of powerful computational tools. First, many clustering methods yield different and sometimes conflicting results. Second, single cell expression can be used to detect previously unrecognized cell types, while the current clustering algorithms are often not sensitive enough to detect novel, sometimes rare, cell types. Third, genomic interactions obtained from bulk samples and single cells are likely to be complementary to each other and reflect different aspects in terms of gene regulation, co-expression and protein-protein interactions. Novel integrative methods are desired to maximize the information we gain from these genomic datasets. To address these challenges, we will develop computational approaches for single cell data analysis. Specifically, we will develop an iterative clustering method for single cell gene expression analysis (Aim 1). Our approach is designed to be robust and sensitive. We will then develop a method to determine active regulatory networks by integrating single cell RNA-Seq dataset and ATAC-Seq from bulk samples (Aim 2). This method will enable us to identify the regulatory circuits at the single cell level. We will then perform single cell RNA-Seq in retinal degenerative models (Aim 3) and apply our computational approaches to the dataset. We expect to identify the drivers and pathways involved in photoreceptor degeneration. Finally, we will develop a database for cell marker genes by analyzing publically available single cell datasets (Aim 4). We believe that the computational algorithms and database we propose to develop will be valuable resource for the research community. The in- depth study on retinal degenerative models will reveal key molecular events that lead to the disease and provide novel therapeutic targets for the retinal degenerative diseases.