r - 每个集群的 FeatureScatter 的 facet_grid

Question

我想知道是否可以为函数facet_grid中的每个集群绘制一个图。FeatureScatter我要绘制的两个特征由 feature1 和 feature2 定义，构面应该是我的对象中定义为pbmc.big$seurat_clusters.

FeatureScatter(object = pbmc.big,
               feature1 = "MALAT1",
               feature2 = "percent.mito",
               plot.cor = T)

Answer 1

值得一提的是，您使用的数据可以从此链接下载并创建如下：

library(Seurat)
library(magrittr)

pbmc.big <- Read10X(data.dir = "../data/pbmc3k/filtered_gene_bc_matrices/hg19/")
pbmc.big <- CreateSeuratObject(counts = pbmc.big)
pbmc.big$percent.mito = PercentageFeatureSet(pbmc.big,pattern="^MT-")

运行集群：

pbmc.big = pbmc.big %>% 
SCTransform() %>% 
RunPCA() %>% 
RunTSNE(dims=1:15) %>% 
FindNeighbors(dims=1:15) %>% 
FindClusters(res=0.1)

你可以这样刻面：

g = FeatureScatter(object = pbmc.big,
               feature1 = "MALAT1",
               feature2 = "percent.mito",
               plot.cor = TRUE)

g + facet_wrap(~colors)

缺少相关性，如果需要，一种方法是提取变量并绘制图：

library(ggpubr)

data.frame(cluster = Idents(pbmc.big),
MALAT1 = FetchData(pbmc.big,"MALAT1"),
percent.mito = FetchData(pbmc.big,"percent.mito")) %>%
ggplot(aes(x=MALAT1 , y=  percent.mito, col = cluster)) +
geom_point(size=1) +
facet_wrap(~cluster)+
stat_cor(method = "pearson")