樣本量比較大的時候推薦使用百分位數(shù)進(jìn)行細(xì)胞過濾

#百分位數(shù)法：
  ncount_q   <- quantile(sample.tmp.seurat@meta.data$nCount_RNA,c(0.025,0.975))
  nfeature_q <- quantile(sample.tmp.seurat@meta.data$nFeature_RNA,c(0.025,0.975))
  mt_q       <- quantile(sample.tmp.seurat@meta.data$percent.mt,c(0.025,0.8))

可視化

pdf(paste(output,"Quality_Control_Count_Feature.pdf",sep="_"), width=8, height=6)
    p3       <- plot(sample.tmp.seurat@meta.data$nCount_RNA,sample.tmp.seurat@meta.data$nFeature_RNA,pch=16,cex=0.7,bty="n")
    p3       <- p3 + abline(h=c(as.numeric(nfeature_q)[1],as.numeric(nfeature_q)[2]),v=c(as.numeric(ncount_q)[1],as.numeric(ncount_q)[2]),lty=2,lwd=2,col="red")
    print(p3)
  dev.off()

可視化后，可以看這個標(biāo)準(zhǔn)過大概過濾掉了多少細(xì)胞

創(chuàng)建過濾后的子集

sample.tmp.seurat <-subset(sample.tmp.seurat,subset=
nFeature_RNA>as.numeric(nfeature_q)[1] & 
nFeature_RNA<as.numeric(nfeature_q)[2]&
nCount_RNA>as.numeric(ncount_q)[1]&
nCount_RNA<as.numeric(ncount_q)[2]&percent.mt<10)