Integrative analysis of spatial transcriptomics with histology images and single cells

项目摘要 的 功能 相对 疾病 我们体内的组织由不同的细胞类型组成，每种细胞类型都专门执行特定的功能。 细胞的行为受其在组织内的周围环境的影响。知识 组织中不同细胞的位置对于理解细胞类型的空间组织是至关重要的， 尽管单细胞RNA测序（scRNA-seq）已经使得表征细胞凋亡成为可能。 类型和状态以前所未有的分辨率，细胞之间缺乏物理关系阻碍了 组织环境中细胞间通讯的研究。空间转录组学研究进展 (ST)能够进行基因表达谱分析，同时保留组织中的位置信息。受欢迎的ST 该技术是基于空间条形码，然后是下一代测序，其中转录组范围的 基因表达在空间条形码点中测量。来自这种ST技术的数据通常包括高- 组织切片的高分辨率苏木精和伊红（H & E）染色的组织学图像， 获得表达数据。尽管ST功能强大，但生成此类数据的成本仍然很高。另 另一方面，生成H & E染色的组织学图像和scRNA-seq数据相对便宜。主要 本项目的动机是利用ST中的信息，从相对简单的 以获得组织学图像和scRNA-seq数据。基于我们在统计基因组学方面的专业知识，我们建议 开发新的机器学习方法，以解决执行任务时的关键计算挑战。 ST、组织学图像和单细胞的综合分析。我们的方法将共同模拟基因表达 和组织学来表征组织的空间组织并预测空间基因表达， 组织学图像。从这些分析中产生的空间地图将进一步使单一的空间映射成为可能。 细胞转化为组织所提出的方法将适用于公共数据和数据产生的持续 在各种疾病的合作，以评估他们的表现。该项目的成功完成将 允许研究人员利用先进的机器学习算法来整合ST，组织学， 单细胞数据，以获得组织空间组织的整体视图。

PROJECT SUMMARY The function. relative disease tissues in our body consist of diverse cell t ypes with each cell type specialized to carry out a particular The behavior of a cell is influenced by its surrounding environment within a tissue. Knowledge of the locations of different cells in a tissue is critical for understanding the spatial organization of cell types and pathology.Although single-cell RNA sequencing (scRNA-seq) has made it possible to characterize cell types and states at an unprecedented resolution, the lack of physical relationships among cells has hindered the study of cell-cell communications within tissue context. Recent technology advances in spatial transcriptomics (ST) have enabled gene expression profiling while retaining location information in tissues. A popular ST technology is based on spatial barcoding followed by next-generation sequencing in which transcriptome-wide gene expression is measured in spatially barcoded spots. Data from such ST technologies often include a high- resolution hematoxylin and eosin (H&E)-stained histology image of the tissue section from which the gene expression data are obtained. Although ST is powerful, such data are still expensive to generate. On the other hand, it is relatively cheaper to generate H&E-stained histology images and scRNA-seq data. The main motivation of this project is to leverage information in ST to gain additional knowledge from the relatively easy- to-obtain histology images and scRNA-seq data. Building upon our expertise in statistical genomics, we propose to develop novel machine learning methods to address key computational challenges when performing integrative analysis of ST, histology images, and single cells. Our methods will jointly model gene expression and histology to characterize the spatial organization of tissues and predict spatial gene expression from histology images. The resulting spatial map from these analyses will further enable the spatial mapping of single cells back to tissues. The proposed methods will be applied to public data and data generated from ongoing collaborations in various diseases to evaluate their performance. The successful completion of this project will allow researchers to take advantage of advanced machine learning algorithms to integrate ST, histology, and single-cell data to gain a holistic view of the spatial organization of tissues.