单细胞测序生信分析2-Seurat

接上一篇的数据准备阶段，上一篇已经介绍了4种情况:
1.有直接的标准10X数据（barcodes.tsv/genes.tsv/matrix.mtx）
2.多个10X数据合并
3.只有矩阵数据
4.矩阵数据还需要筛选

最近又学习了几种情况，再来补充下：
5.多个csv/txt矩阵合并：

# 读取多个csv
data1 <- read.csv("******", header = T,row.names= 1)
data2 <- read.csv("******", header = T,row.names= 1)
data3 <- read.csv("******", header = T,row.names= 1)

#如果是txt文件，就用data = read.table("******", header = T,row.names= 1)

# 合并成数据框
datan = data.frame(data1,data2,data3)

# 数据框转换成稀疏矩阵matrix
dataan <- as(as.matrix(datan), "dgCMatrix")

6.数据分组比较多，想要R实现批量处理
例如：GSE162025

image.png

library(data.table)
samples=list.files('GSE162025_RAW/')
samples 
pbmcList = lapply(samples,function(x){ 
  x=samples[1]
  print(x)
  y=file.path('GSE162025_RAW',x )  
  a=fread(y,data.table = F)
  a[1:4,1:4] 
  rownames(a)=a[,1]
  a=a[,-1]
  pbmc=CreateSeuratObject(a)
  return(pbmc)
})
pbmcList

pbmc=merge(x=pbmcList[[1]],
           y=pbmcList[ -1 ])
as.data.frame(pbmc@assays$RNA@counts[1:10, 1:2])
head(pbmc@meta.data)

library(stringr)
phe=str_split(rownames(pbmc@meta.data),'_',simplify = T)
head(phe)
pbmc@meta.data$patient=phe[,2]
pbmc@meta.data$location=phe[,3]
pbmc@meta.data$orig.ident=paste(phe[,3],phe[,2],sep = '_')
table(pbmc@meta.data) 

这个代码是来自Jimmy老师，运行比较慢

如果想CCA合并，运行就更慢了

#保证当前工作目录下面后缀是matrices.csv.gz
options(stringsAsFactors = F)

fs=list.files(pattern = 'matrices.csv.gz')
fs

pbmcList <-  lapply(fs, function(x){
  a=read.csv( x )
  a[1:4,1:4]
  raw.data=a[,-1]
  rownames(raw.data)=a[,1]
  library(stringr)
  p=str_split(x,'_',simplify = T)[,2]
  pbmc <- CreateSeuratObject(counts = raw.data,project = p )
})

#多个单细胞对象的整合
pro='integrated' 

for (i in 1:length(pbmcList)) {
  pbmcList[[i]] <- NormalizeData(pbmcList[[i]], verbose = FALSE)
  pbmcList[[i]] <- FindVariableFeatures(pbmcList[[i]], selection.method = "vst", 
                                       nfeatures = 2000, verbose = FALSE)
}
pbmcList
pbmc.anchors <- FindIntegrationAnchors(object.list = pbmcList, dims = 1:30)
pbmc.integrated <- IntegrateData(anchorset = pbmc.anchors, dims = 1:30)

DefaultAssay(pbmc.integrated) <- "integrated"

head(pbmc.integrated@meta.data)
tail(pbmc.integrated@meta.data)