Methods for inferring and analyzing gene regulatory networks using single-cell multiomics and spatial genomics data

项目总结/摘要 基因调控的研究提供了基本的机制知识，以破译的发展， 复杂的生物系统并通过操纵细胞状态来设计疾病的治疗策略。发展 单细胞多模式和多组学测序技术的发展， 信息，如顺式调控元件，它使基因调控的推断超越了简单的基因， 共表达然而，为了更准确地识别基因调控，迫切需要计算方法 将在同一细胞中测量的转录组学和表观基因组学数据结合起来。近年来研究表明 表观基因组学的变化可能预示着基因转录，表明基因调控是异步的， 过程一组调节子可能不会沿着整个连续的生物学过程而影响靶基因， 只在很短的时间内然而，从时间序列数据推断的当前基因调控网络提供了 平均监管时间，而不是表征监管动态。此外，先前的证据表明， 细胞间的通讯和相互作用在空间定位的细胞群中以规则的形式存在， 化的活动。进一步将随时间变化的发展轨迹纳入组织空间景观将有助于 对基因调控的空间异质性进行分类。然而，我们仍然缺乏计算方法来计算- 在基因调控推断中的时空轨迹。为了缩小这些差距，我的研究计划将 (1)发展计算方法和统计方法来推断静态和动态基因调控 网络，（2）利用从多组学和空间转录组学数据推断的空间轨迹来表征， 空间基因调控，以及（3）开发基因调控网络的控制策略，以实现所需的 使用多样本单细胞多组学数据的状态。我的长期目标是描述基因调控的特征， 多个样品和在空间上分离的组织区域中，并鉴定用于manip的驱动基因的最小集合。 评估小区状态（例如，细胞类型）。基于我以前在基因调控网络推理方面的经验， 方法开发分析多样本单细胞基因组学和表观基因组学数据，我将开发新的 使用多样本单细胞多组学数据和特征来推断基因调控网络的统计方法， 时间滞后的活动。我将建立在我最近的工作单细胞伪时间分析和单细胞 时空轨迹来推断时间和时空背景中的基因调控。此外，建筑 在多样品单细胞分析的网络控制和方法开发方面的理论研究 组学数据，我将开发基因调控网络的控制策略，以实现所需的细胞状态或细胞凋亡。 最小化驱动基因集的进化轨迹。最后，我将系统地评价这些方法 在组织和样品条件下使用公开可用的数据。所有方法将普遍适用于 不同的组织或疾病背景。这些方法将向社会公开和免费发布， 可用的开源软件，可以使广大的科学界受益。

Project Summary/Abstract The study of gene regulation provides fundamental mechanistic knowledge to decipher the development of com- plex biological systems and design treatment strategies for diseases by manipulating cell states. Development of single-cell multimodal and multiomics sequencing technologies allows for profiling of additional epigenomics information, such as cis-regulatory elements, which enables inference of gene regulation beyond simply gene coexpression. However, to more accurately identify gene regulation, computational methods are urgently needed to incorporate both transcriptomics and epigenomics data measured in the same cell. Recent studies show that epigenomics changes may foreshadow gene transcription, indicating that gene regulation is an asynchronous process. A set of regulators may not influence the target gene along the whole continuous biological process but only in a small time window. However, current gene regulatory networks inferred from time series data provide average regulation across time rather than characterizing regulation dynamics. Also, prior evidence shows that cell–cell communication and interaction are present in spatially localized cell populations as a form of regular- ized activity. Further incorporating time-varying developmental trajectories in tissue-spatial landscapes will help categorize the spatial heterogeneity of gene regulation. However, we still lack computational methods to incor- porate spatio-temporal trajectories in gene regulation inference. To close these gaps, my research program will (1) develop computational methods and statistical approaches to infer both static and dynamic gene regulatory networks, (2) leverage the spatial trajectories inferred from multiomics and spatial transcriptomics data to charac- terize spatial gene regulation, and (3) develop control strategies for gene regulatory networks to achieve desired states using multi-sample single-cell multiomics data. My long-term goal is to characterize gene regulation across multiple samples and in spatially separated tissue regions and identify the minimum set of driver genes to manip- ulate cell states (e.g., cell types). Building on my previous experience in gene regulatory network inference and methods development to analyze multi-sample single-cell genomics and epigenomics data, I will develop novel statistical methods to infer gene regulatory networks using multi-sample single-cell multiomics data and char- acterize time-lagged activities. I will build on my recent work on single-cell pseudotime analysis and single-cell spatio-temporal trajectories to infer gene regulation in temporal and spatio-temporal contexts. Further, building on my theoretical research in network control and methods development in analyzing multi-sample single-cell omics data, I will develop control strategies for gene regulatory networks to achieve desired cell states or cell de- velopmental trajectories with a minimum set of driver genes. Finally, I will systematically evaluate these methods across tissues and sample conditions using publicly available data. All methods will be generally applicable in different tissue or disease contexts. These methods will be released to the community with publicly and freely available open-source software that can benefit the broad scientific community.