Statistical Power Calculation Framework for Spatially Resolved Transcriptomics Experiments

摘要 最近，高通量空间转录组学(HST)技术(例如，10倍基因组学维西姆，幻灯片序列， 和Slide-seqV2)使同时测量接近细胞水平的基因表达和 这些细胞在组织或器官内的空间位置。这些新技术提供了前所未有的 研究细胞异质性和细胞间通讯的机会。虽然有一些计算量 最近出现了用于HST数据分析的工具，这是HST设计的一个严格的统计框架 文献中仍然缺少实验。计划进行HST实验的研究人员需要确定各种 实验设计参数，如测序深度，以及这些选择是否影响关键目标 可以实现HST实验，例如，鉴定组织结构、空间可变基因和细胞- 手机通讯。在这项提案中，我们的目标是为HST开发一个严格的功率分析框架 实验。组建的团队在HST数据的统计建模方面拥有强大和互补的专业知识， 高通量测序功率分析和设计的统计框架和软件的开发 数据、单细胞基因组学技术、空间统计、计算工具开发和利用 这些计算工具用于研究疾病的分子和免疫学基础。我们将实现 通过实施两个具体目标来提出目标。在目标1中，我们将开发一个严格的功率分析框架 用于HST实验。在目标2中，我们将开发一个交互式Web界面和一个用于功率分析的R包 HST实验的结果。建议的功率分析框架将通过模拟进行开发和评估 数据、公共领域的HST数据以及来自合作者的内部HST数据集。统计框架 将在该项目中开发的，以及实现该框架的开源软件，将 为未来HST实验的优化设计提供必要的工具。

Abstract Recently, high-throughput spatial transcriptomics (HST) technologies (e.g., 10X Genomics Visium, Slide-seq, and Slide-seqV2) have made it possible to simultaneously measure close-to-cell-level gene expressions and spatial locations of these cells within a tissue or organ. These new technologies have provided an unprecedented opportunity to investigate cellular heterogeneity and cell-cell communications. Although a few computational tools for HST data analysis have recently become available, a rigorous statistical framework for design of HST experiments is still missing in the literature. Researchers planning an HST experiment need to determine various experimental design parameters such as the sequencing depth, and these choices affect whether key goals of HST experiments can be achieved, e.g., identification of tissue architecture, spatially variable genes, and cell- cell communications. In this proposal, we aim to develop a rigorous power analysis framework for HST experiments. The assembled team has strong and complementary expertise in statistical modeling of HST data, development of statistical frameworks and software for power analysis and design of high throughput sequencing data, single-cell genomics technologies, spatial statistics, computational tool development, and utilization of these computational tools for investigation of molecular and immunologic basis of diseases. We will achieve the proposed goal by implementing two specific aims. In Aim 1, we will develop a rigorous power analysis framework for HST experiments. In Aim 2, we will develop an interactive web interface and an R package for power analysis of HST experiments. The proposed power analysis framework will be developed and evaluated using simulation data, HST data in the public domain, and in-house HST datasets from collaborators. The statistical framework that will be developed in this project, along with the open-source software implementing this framework, will provide essential tools for the optimal design of future HST experiments.