其他
TCGA批量差异分析并可视化
Try to learn everyhting about something!
生信技能树之前推过一篇推文：多次差异分析难道就需要多个火山图吗？
里面介绍过一种使用条形图展示多组差异分析结果。
今天学习下这个图的画法，不过不是多组差异分析的结果，而是多个癌症的差异分析结果，就是用TCGA的33种癌症。
差异分析
首先读取33个癌症的raw counts表达矩阵，这个是从xena网站下载的，简单处理下即可。生息技能树也有教程的，搜索即可获得！
读取数据
# 清空变量
rm(list = ls())
# 加载R包
library(tinyarray)
# 获取路径
fs <- list.files('Rdata/',pattern = 'htseq_counts')
fs
[1] "TCGA-ACC.htseq_counts.Rdata"
[2] "TCGA-BLCA.htseq_counts.Rdata"
[3] "TCGA-BRCA.htseq_counts.Rdata"
[4] "TCGA-CESC.htseq_counts.Rdata"
[5] "TCGA-CHOL.htseq_counts.Rdata"
[6] "TCGA-COAD.htseq_counts.Rdata"
[7] "TCGA-DLBC.htseq_counts.Rdata"
[8] "TCGA-ESCA.htseq_counts.Rdata"
[9] "TCGA-GBM.htseq_counts.Rdata"
[10] "TCGA-HNSC.htseq_counts.Rdata"
[11] "TCGA-KICH.htseq_counts.Rdata"
[12] "TCGA-KIRC.htseq_counts.Rdata"
[13] "TCGA-KIRP.htseq_counts.Rdata"
[14] "TCGA-LAML.htseq_counts.Rdata"
[15] "TCGA-LGG.htseq_counts.Rdata"
[16] "TCGA-LIHC.htseq_counts.Rdata"
[17] "TCGA-LUAD.htseq_counts.Rdata"
[18] "TCGA-LUSC.htseq_counts.Rdata"
[19] "TCGA-MESO.htseq_counts.Rdata"
[20] "TCGA-OV.htseq_counts.Rdata"
[21] "TCGA-PAAD.htseq_counts.Rdata"
[22] "TCGA-PCPG.htseq_counts.Rdata"
[23] "TCGA-PRAD.htseq_counts.Rdata"
[24] "TCGA-READ.htseq_counts.Rdata"
[25] "TCGA-SARC.htseq_counts.Rdata"
[26] "TCGA-SKCM.htseq_counts.Rdata"
[27] "TCGA-STAD.htseq_counts.Rdata"
[28] "TCGA-TGCT.htseq_counts.Rdata"
[29] "TCGA-THCA.htseq_counts.Rdata"
[30] "TCGA-THYM.htseq_counts.Rdata"
[31] "TCGA-UCEC.htseq_counts.Rdata"
[32] "TCGA-UCS.htseq_counts.Rdata"
[33] "TCGA-UVM.htseq_counts.Rdata"
# 批量读取
dat_list <- lapply(fs, function(x){
pro <- gsub('.htseq_counts.Rdata','',x)
load(file = file.path('Rdata/',x))
dat <- pd_mat
return(dat)
})
# 获取名称
types <- gsub(".htseq_counts.Rdata","",fs)
types
[1] "TCGA-ACC" "TCGA-BLCA" "TCGA-BRCA" "TCGA-CESC"
[5] "TCGA-CHOL" "TCGA-COAD" "TCGA-DLBC" "TCGA-ESCA"
[9] "TCGA-GBM" "TCGA-HNSC" "TCGA-KICH" "TCGA-KIRC"
[13] "TCGA-KIRP" "TCGA-LAML" "TCGA-LGG" "TCGA-LIHC"
[17] "TCGA-LUAD" "TCGA-LUSC" "TCGA-MESO" "TCGA-OV"
[21] "TCGA-PAAD" "TCGA-PCPG" "TCGA-PRAD" "TCGA-READ"
[25] "TCGA-SARC" "TCGA-SKCM" "TCGA-STAD" "TCGA-TGCT"
[29] "TCGA-THCA" "TCGA-THYM" "TCGA-UCEC" "TCGA-UCS"
[33] "TCGA-UVM"
批量对33种癌症进行差异分析
使用DESeq2进行差异分析，真的是3个差异分析R包中最简单的一个了，直接一句代码即可！
library(DESeq2)
for(i in 1:length(dat_list)){
group <- make_tcga_group(dat_list[[i]])
if(length(unique(group)) <2) next # 有的癌症没有正常样本，不要
coldata <- data.frame(sample = colnames(dat_list[[i]]),
group = group)
dds <- DESeqDataSetFromMatrix(
countData = dat_list[[i]],
colData = coldata,
design = ~ group
)
dds <- DESeq(dds) # deseq2用起来是真的简单
res <- results(dds,tidy = T)
res2 <- res[abs(res$log2FoldChange) > 1 & res$padj < 0.05,] # 提取差异基因
save(res2, file = paste0(types[i],"_diff_genes.Rdata"))
}
最终获得了24个癌症的差异基因。
把数据都保存好，接下来就可以画图了。
多个差异分析结果的条形图
构造画图数据
# 读取数据
rm(list = ls())
files <- list.files("./", pattern = ".Rdata")
alldiff <- lapply(files, function(x){
load(file = x)
dat <- res2
return(dat)
})
names(alldiff) <- gsub("-diff.Rdata","",files)
alldiff <- lapply(alldiff,na.omit) # 去掉NA
数就是图，图就是数，所以最重要的就是构造数据。
df_dim <- lapply(alldiff, function(x){
dime <- cut(x$log2FoldChange,
breaks = c(-Inf,-4,-3,-2.5,-2,-1.5,-1,1,1.5,2,2.5,3,4,Inf)) # 切割值
return(data.frame(table(dime)))
})
library(dplyr)
df <- bind_rows(df_dim) # 列表变成数据框
df$type <- rep(names(alldiff),each=13) # 添加癌症分类
str(df)
'data.frame': 312 obs. of 3 variables:
$ dime: Factor w/ 13 levels "(-Inf,-4]","(-4,-3]",..: 1 2 3 4 5 6 7 8 9 10 ...
$ Freq: int 80 240 211 311 454 742 0 960 576 401 ...
$ type: chr "TCGA-BLCA" "TCGA-BLCA" "TCGA-BLCA" "TCGA-BLCA" ...
df就长这样，整洁的长数据，适合画图：
画图
library(ggplot2)
ggplot(df, aes(x=type, y=Freq, fill=dime))+
geom_bar(stat = "identity")+
geom_text(aes(label = Freq),position = position_stack(vjust = 0.5),size=2)
第一个图长这样，不太好看的样子...下面美化一下。
美化一下：
df_sum <- df %>% group_by(type) %>% summarise(n=sum(Freq)) # 计算每个癌症总的差异基因数目
# 构造色盘
library(RColorBrewer)
my_col <- c(rev(brewer.pal(6,"Reds")),"white",brewer.pal(6,"Blues"))
# 画图
ggplot()+
geom_bar(data = df,mapping = aes(x=type, y=Freq, fill=dime),stat = "identity")+
geom_text(data = df_sum,mapping = aes(x=type,y=n,label=n),vjust=-1)+
scale_fill_manual(values = my_col,name = NULL)+
scale_y_continuous(expand = c(0,0),name = "Counts",limits = c(0,8000))+
theme_minimal()+
theme(axis.text.x = element_text(angle = 45,vjust = 1,hjust = 1),
axis.title.x = element_blank())
这样看起来还是不错的！
但是这个图有什么用呢？
以上就是今天的内容，希望对你有帮助哦！欢迎点赞、在看、关注、转发！
欢迎在评论区留言或直接添加我的微信！
欢迎关注公众号：医学和生信笔记
“医学和生信笔记 公众号主要分享：1.医学小知识、肛肠科小知识；2.R语言和Python相关的数据分析、可视化、机器学习等；3.生物信息学学习资料和自己的学习笔记！
往期回顾
R语言画好看的聚类树
生信图学习之网络图
ggtern画三元图