TCGA转录组数据
转录组数据的差异分析与芯片数据的差异分析有不同，这里我们使用count数据，因为这几种包中包装了标准化的函数，只需要程序化的输入两个数据，count矩阵和group数据，就能得到差异分析的结果。那么三种差异分析的包有什么差异呢？
一、数据整理
这里我们需要count矩阵，以及对应的分组信息
rm(list = ls())
library(tidyverse)
library(limma)
library(stringr)
library(xlsx)
library(edgeR)
load("count_exp.Rdata")
cli <- read.xlsx2("../2.clinical/cli.xlsx",sheetIndex = 2)
#-------------------------------------------------------------------------
group <- cli[,c(1,10)]
dat <- count_exp[,str_sub(colnames(count_exp),1,15)%in%group$Sample.ID]
group <- group[group$Sample.ID%in%str_sub(colnames(count_exp),1,15),]
dat <- dat[,match(group$Sample.ID,str_sub(colnames(dat),1,15))] %>%
as.data.frame()
Group <- factor(rep(c("re","non_re"),times=c(26,47)),levels = c("non_re","re"))
table(Group)
二、limma包
#-----limma------------------------------------------------------------------
design <- model.matrix(~0+Group)
colnames(design)= levels(Group)
rownames(design)=colnames(dat)
dge <- DGEList(counts=dat)
dge <- calcNormFactors(dge)
v <- voom(dge,design, normalize="quantile")
fit <- lmFit(v, design)
constrasts = paste(rev(levels(Group)),collapse = "-")
cont.matrix <- makeContrasts(contrasts=constrasts,levels = design)
fit2=contrasts.fit(fit,cont.matrix)
fit2=eBayes(fit2)
DEG = topTable(fit2, coef=constrasts, n=Inf)
DEG = na.omit(DEG)
logFC_t=1
P.Value_t = 0.05
k1 = (DEG$P.Value < P.Value_t)&(DEG$logFC < -logFC_t)
k2 = (DEG$P.Value < P.Value_t)&(DEG$logFC > logFC_t)
change = ifelse(k1,"down",ifelse(k2,"up","stable"))
DEG$change <- change
DEG_limma <- DEG
save(DEG_limma,group,file = "DEG_limma.Rdata")
二、DESeq2包
#--DESeq2----------------------------------------------------------------------------
library(DESeq2)
dat <- apply(dat,2,as.integer) #整数转换
rownames(dat) <- rownames(count_exp)
colData <- data.frame(row.names =colnames(dat),
condition=Group)
dds <- DESeqDataSetFromMatrix(
countData = dat,
colData = colData,
design = ~ condition)
dds <- DESeq(dds)
res <- results(dds, contrast = c("condition",rev(levels(Group))))
resOrdered <- res[order(res$pvalue),] # 按照P值排序
DEG <- as.data.frame(resOrdered)
DEG = na.omit(DEG)
head(DEG)
logFC_t=1
P.Value_t = 0.05
k1 = (DEG$padj < P.Value_t)&(DEG$log2FoldChange < -logFC_t)
k2 = (DEG$padj < P.Value_t)&(DEG$log2FoldChange > logFC_t)
change = ifelse(k1,"down",ifelse(k2,"up","stable"))
DEG$change <- change
#down stable up
#131 17964 40
DEG_DEseq2 <- DEG
save(DEG_DEseq2,group,file = "DEG_DEseq2.Rdata")
三、edegR包
#---edgeR------------------------------------------------------------------
library(edgeR)
dge <- DGEList(counts=dat,group=Group)
dge$samples$lib.size <- colSums(dge$counts)
dge <- calcNormFactors(dge)
design <- model.matrix(~0+Group)
rownames(design)<-colnames(dge)
colnames(design)<-levels(Group)
dge <- estimateGLMCommonDisp(dge, design)
dge <- estimateGLMTrendedDisp(dge, design)
dge <- estimateGLMTagwiseDisp(dge, design)
fit <- glmFit(dge, design)
fit2 <- glmLRT(fit, contrast=c(-1,1))
DEG=topTags(fit2, n=nrow(exp))
DEG=as.data.frame(DEG)
logFC_t=1
P.Value_t = 0.05
k1 = (DEG$PValue < P.Value_t)&(DEG$logFC < -logFC_t)
k2 = (DEG$PValue < P.Value_t)&(DEG$logFC > logFC_t)
change = ifelse(k1,"down",ifelse(k2,"up","stable"))
DEG$change <- change
#down stable up
# 891 18206 408
DEG_edgeR <- DEG
save(DEG_edgeR,group,file = "DEG_edgeR.Rdata")
看看三个包的基因交叉情况
这里，我们使用upset图进行绘制
代码如下：
rm(list = ls())
load("DEG_limma.Rdata")
load("DEG_DEseq2.Rdata")
load("DEG_edgeR.Rdata")
UP=function(df){
rownames(df)[df$change=="up"]
}
DOWN=function(df){
rownames(df)[df$change=="down"]
}
DIFF=function(df){
rownames(df)[df$change != "stable"]
}
up = intersect(intersect(UP(DEG_DEseq2),UP(DEG_edgeR)),UP(DEG_limma))
down = intersect(intersect(DOWN(DEG_DEseq2),DOWN(DEG_edgeR)),DOWN(DEG_limma))
#上调、下调基因分别画维恩图
up_genes = list(UP(DEG_DEseq2),
UP(DEG_edgeR),
UP(DEG_limma))
down_genes = list(DOWN(DEG_DEseq2),
DOWN(DEG_edgeR),
DOWN(DEG_limma))
library(gplots)
venn(up_genes,show.plot = T)
venn(down_genes)
#汇集信息，绘制upset图
library(UpSetR)
input <- c("DEG_DEseq2"=171,
"DEG_edgeR"=1299,
"DEG_limma"=178,
"DEG_limma&DEG_edgeR"=121,
"DEG_limma&DEG_DEseq2"=47,
"DEG_DEseq2&DEG_edgeR"=141,
"DEG_limma&DEG_edgeR&DEG_DEseq2"=44)
data <- fromExpression(input)
upset(data)
upset
最后
可以看到，三个包得到的差异基因还是有差异的，在在这里，我的样本数目较少，使用的TCGA晚期复发-未晚期复发的数据。