各位同学，大家好，今天我们来分享一篇新的10X单细胞和10X空间转录组联合分析文章，文章在这里Spatiotemporal single-cell RNA sequencing 1 of developing hearts reveals interplay between cellular differentiation and morphogenesis,里面有很多经典的10X单细胞和10X空间转录组的分析方法，今天我们的任务就是来参透这篇文章。
一、ABSTRACT
Single-cell RNA sequencing is a powerful tool to study developmental biology but does not preserve spatial information about cellular interactions and tissue morphology（单细胞无法提供细胞的空间位置信息和组织形态学信息，这个已经是单细胞很明显的劣势），Here, we combined single-cell and spatial transcriptomics with new algorithms for data integration to study the early development of the chicken heart.（单细胞空间联合来分析鸡心脏的发育过程，这个很新颖，别说单细胞空间联合的文章，光单细胞的文章研究特殊物种的都很少），We collected data from four key ventricular（心室的） development stages, ranging from the early chamber formation stage to the late four chambered stage.（取样，相对很好取，鸡嘛，比人的样本要容易），We created an atlas of the diverse cellular lineages in developing hearts, their spatial organization, and their interactions during development. Spatial mapping of differentiation transitions revealed the intricate interplay between cellular differentiation and morphogenesis in cardiac cellular lineages（得到的结果，跟预想差不多）。Using spatially resolved expression analysis, we identified anatomically restricted gene expression programs. Last, we discovered a stage dependent role for the small secreted peptide, thymosin beta-4, in the coordination of multi lineage cellular populations. Overall, our study identifies key stage-specific regulatory programs that govern cardiac development.（看来细胞的空间位置，那是相当重要）。
二、INTRODUCTION
这部分主要说了一下几方面的内容
1、运用10X单细胞空间两种技术，we combined spatially resolved RNA-seq with high throughput scRNA-seq to study the spatiotemporal interactions and regulatory programs that drive fetal development of the chicken heart。（时空分析，这个看过我文章分析的同学应该不陌生）。Current spatial transcriptomics approaches lack single-cell resolution, which we addressed here using new approaches to integrate high throughput spatial and single-cell transcriptomic data.（两种分析方法强强联合）。
2、为什么选择鸡的心脏样本。the chick heart anatomy resembles the anatomy of the human heart more closely than other non-mammalian vertebrate model organisms（鸡的心脏结构跟人的很接近）。取鸡心脏的样本来运用单细胞和空间技术。
3、两种技术联合分析的效果。As we demonstrate here, the combination of single-cell and spatial transcriptomics uniquely enables to unravel cellular interactions that drive cardiogenesis.The data enabled us to reconstruct a high-resolution, spatially resolved gene expression atlas of epi-, endo-, and myocardial developmental lineages within cardiac tissue.We used the cell-type predictions to construct proximity maps revealing novel cellular interactions。我们注意一点，We furthermore constructed a similarity map between single-cell and spatial transcriptomes, which enabled us to spatially map lineage-associated differentiation trajectories within the tissue.（单细胞轨迹分析在空间上的反映，这个据我所知，应该是首次）。
三、Result
（1）Spatially resolved single-cell transcriptomics atlas of developing fetal chicken hearts
这部分结果主要讲了两种技术的联合运用。To study the complex interplay between differentiation and morphogenesis during cardiac development, we combined single-cell and spatial transcriptomics.看一下技术路线
不同时间点取样，同时进行10X单细胞和10X空间转录组技术，然后单细胞空间的联合分析，最后是一些个性化，包括细胞之间的相互作用，轨迹分析等等。
spatial transcriptomes collected at the same developmental stage were strongly correlated（Pearson correlation; R > 0.98）。同一时间段的单细胞和空间样本具有极强的关联性。To analyze single-cell transcriptomes, we filtered and preprocessed the data, performed batch correction using scanorama（单细胞样本之间的批次去除采用了scanorama的方法，问一下大家这个方法是哪个软件的？知道的写在评论区哈，作者是知道的），我们看一下单细胞的分析结果。
鸡心脏的细胞定义，相信非常的难，我们国内的学者大部分做细胞定义完全没有系统，可能这就是国内外科研的差距吧，关键这细胞定义还是用marker gene定义的，这个定义过程及科研素质，令人敬佩。
接下来是空间数据的分析，populations, the spatial transcriptomics data was integrated with the scRNA-seq data using Seurat-v3 anchor-based integration（Seurat的联合方法，这个方法我之前介绍过，文章在10X空间转录组和10X单细胞数据联合分析方法汇总），
In order to understand the spatial organization of cell types in broad anatomical regions, spots were labeled as cell types with maximum prediction score and visualized on H&E stained images of respective stages（每个SPOT定义成联合分析中分数最大的那种细胞类型，如下图)。
Cell-type prediction scores for spatial transcriptomes were further used to estimate the abundance of pairs of specific cell types（细胞共定位），As proximity is a necessity for physical interactions between two or more cells, these cell-type proximity maps can be used to guide the discovery of interactions between cell types from the same or different lineages.
（这个地方也是一个研究的重点，细胞之间的空间位置结果会体现出细胞之间的相互作用）。如下图：细胞之间的定位关系
We found a significant number of cardiomyocytes colocalized with myocardial progenitor cells and precursor cells in all stages, as expected.（细胞之间的空间位置关系）。当然，不同阶段的细胞定位关系有所不同。
（2）Spatially resolved cardiac lineage analysis
Ventricular development in fetal hearts involves regulatory interactions and the coordinated
migration of cells from multiple lineages to form a fully developed four-chambered heart.（任何组织的发育都有时空的关系），
We first gathered and reclustered single-cell transcriptomes from the epicardial, endocardial, and myocardial lineages, and then reconstructed differentiation trajectories for these three lineages.（相同细胞类型的再分群和轨迹分析），We used PHATE (Potential of Heat-diffusion for Affinity-based Transition Embedding) to visualize differentiation trajectories because of its ability to learn and conserve local and global structure in low dimensional space（PHATE方法我之前分享过，这个方法目前来讲降维可视化效果最好，文章在这里10X单细胞降维分析之PHATE）,PHATE1作为发育轨迹点，如下图：
我们来看一下主要的分析结果
当然这个图得到了很多有意义的结果，但是这里作为信息分析人员，有以下分析点需要注意：
（1）、相同细胞类型的再分群。
（2）、PHATE降维可视化
（3）、单细胞空间联合展示不同阶段的细胞类型在空间上的位置。
其中，作者也用了monocle2进行轨迹分析。
（3）Spatiotemporally resolved local cellular heterogeneity in developing cardiac tissue
空间数据聚类，To examine transcriptional differences within the fetal cardiac ventricles,we performed unsupervised clustering of spatial RNA-seq spots and labeled the clusters by anatomical region based on their location in the tissue。（注意看这里，空间聚类的注释采用的是region，这个地方一定要注意哈。）Using this analysis, we identified distinct
spatial clusters derived from ventricles, atria, valves, and the outflow tract but also distinct layers of ventricular regions including epicardium, compact and trabecular myocardium regions, and endocardium。
我们稍微结合一下上面的结果，第一部分的结果是利用单细胞的数据进行空间信息的注释，而这里的结果却是空间单独聚类，进行结构注释，We used cell type prediction scores for spatial transcriptomes as a proxy for cellular composition and analyzed temporal changes in local cellular composition for the major anatomical regions within ventricular tissue（这个地方很有意思，同一区域分析细胞类型成分的时间关系，值得注意，是一个新的思路）。
这个地方主要体现具体区域的细胞异质性（空间marker，SeuratV3寻找空间高变基因）。
（4）A collective cell type and stage dependent role for thymosin beta-4
第四部分的结果我们就不分享了，大家自己看一看，
某一种细胞类型的详细分析。
四、METHOD，这里我们关注一下几种分析方法
（1）、单细胞数据的处理
Scanorama去除批次效应（什么软件？知道的写在评论区里），单细胞分析软件是SeuratV3。
（2）、空间转录组的数据处理，也是SeuratV3，包括单细胞空间联合
（3）、轨迹分析
降维可视化 PHATE，同时进行monocle2分析。
生活很好，有你更好
