空转数据分析之细胞“社区”

作者，Evil Genius

大家好，又到了分享空间转录组分析的时候了，还是那句话，要学我们就好好学习，深入的理解并应用空转，不学就算了，找公司做完全可以，不要那种半学不学的，自己折腾了半天，最后又找公司或者别人了，这种折腾完全没有必要。

还记得空间的四个矩阵么，课上讲过

1、分子矩阵，gene X barcode，这是最开始大家拿到的矩阵 2、细胞矩阵，空间解卷积之后的矩阵，细胞 X Barcode 3、分子niche矩阵， 即分子生态位矩阵，主要研究分子微环境，包括邻域通讯等， gene X Barcode 4、细胞niche矩阵，即细胞生态位矩阵，主要用来研究细胞的空间排布，例如侵袭性的肿瘤细胞空间临近巨噬，所有细胞的空间排布形成了细胞niche矩阵.

这里大家可以参考之前分享的共定位分析

空转第10课共定位内容补充（通路 && 细胞类型）

10X空间转录组数据分析之细胞类型与生物学通路的空间依赖性

10X空间转录组数据分析之细胞的空间依赖性

这里多说一句，即使是高精度平台，例如华大、HD，最后还是要合并分析，逃不出这4个矩阵的范畴。还有寻因的技术，如果真的效果可以，就可以直接拿到细胞矩阵了，不过待考察。

今天分享的空间细胞“社区”分析，主要针对空转的第四个矩阵，细胞niche矩阵，效果图如下

Spatial distribution of cellular neighborhoods

提醒大家一句，个性化分析要根据自己的需求来，而且可以进行延伸，会分析之后思路就会变得非常重要。

包括华大的技术平台也有这样的分析，文章在Identification of HSC/MPP expansion units in fetal liver by single-cell spatiotemporal transcriptomics | Cell Research (nature.com)，文章把这个分析描述成'unit'，其实跟我们今天讲的细胞“社区”一个道理。

Identification of expansion units of HSCs/MPPs To determine the architectural basis of cell–cell interactions, we defined HSC/MPP-localized spots as intra-spots (indicating the closest relationship), HSC/MPP-surrounded spots as inter-spots (indicating the second closest relationship), and other distant spots (indicating nearly no interactive relationship).For each niche cell type, analysis of enrichment score for different types of spots showed that EC with the highest score for intra-spots was close to HSCs/MPPs, which is consistent with a previous report.hepatoblast and stromal cell with the highest scores respectively for inter-spots and other distant spots were less close to HSCs/MPPs; unexpectedly, macrophage with higher scores for intra- and inter-spots than that for other distant spots was considered as a novel niche component spatially close to HSCs/MPPs .To quantitatively compare the interaction between HSCs/MPPs and different niche cells, we defined an enrichment fold based on the ratio of the enrichment score median for each spot type to the enrichment score median for all spots. Consequently, we found that macrophage showed a 11.52-fold enrichment in the intra-spots and a 1.31-fold in the inter-spots, EC showed a 1.62-fold enrichment in the intra-spots, while hepatoblast and stromal cell showed less enrichment in the intra-spots.Furthermore, to validate the spatial relationship at nearly single-cell resolution, we analyzed the mouse E13.5 FL ST data based on Stereo-seq, which is a sequencing-based spatially resolved transcriptomic technology with subcellular resolution.We defined 20 bins as a spot (10–15 μm in diameter), which may include 1–3 cell(s), and then annotated the spots (including

, and other distant ) referring to the aforementioned pipeline. As a result, we found that macrophages showed a high enrichment both in and

.Taken together, the results from three analytic methods of spatial information support that macrophage serve as an important niche cell with the closest relationship with HSCs/MPPs.

先了解一下概念：细胞社区（Cell Neighborhood）是基于距离中心细胞特定距离范围内不同细胞类型的局部密度区分的组织区域。

分析方法呢，也很简单，如下描述

For each cell, the 20 nearest neighbors were determined based on their Euclidean distance of the X and Y coordinates, thereby creating one 'window' of cells per individual cell. Next, we grouped these windows using k-means clustering according to the proportions of cell types within each window. We selected K=11 for the number of neighborhoods as we observed that higher values of k did not result in an improved biologically interpretable number of neighborhoods. Neighborhoods were annotated based on their biological function in normal lymph nodes or their enriched cell type(s)/state(s)

当然其中邻域的数量我们是可以自定义的。

代码在空转数据分析之细胞“社区”

生活很好，有你更好