R语言-分析geo表达谱数据

第一讲：geo，表达芯片和R step0-install.R

rm(list = ls())   
options()$repos 
options()$BioC_mirror
options(BioC_mirror="https://mirrors.ustc.edu.cn/bioc/")
options("repos" = c(CRAN="https://mirrors.tuna.tsinghua.edu.cn/CRAN/"))
options()$repos 
options()$BioC_mirror


# https://bioconductor.org/packages/release/bioc/html/GEOquery.html
if (!requireNamespace("BiocManager", quietly = TRUE))
  install.packages("BiocManager")
BiocManager::install("KEGG.db",ask = F,update = F)
BiocManager::install(c("GSEABase","GSVA","clusterProfiler" ),ask = F,update = F)
BiocManager::install(c("GEOquery","limma","impute" ),ask = F,update = F)
BiocManager::install(c("genefu","org.Hs.eg.db","hgu133plus2.db" ),ask = F,update = F)


# source("https://bioconductor.org/biocLite.R") 
# library('BiocInstaller') 
# options(BioC_mirror="https://mirrors.ustc.edu.cn/bioc/") 
# BiocInstaller::biocLite("GEOquery")
# BiocInstaller::biocLite(c("limma"))
# BiocInstaller::biocLite(c("impute"))

options()$repos
install.packages('WGCNA')
install.packages(c("FactoMineR", "factoextra"))
install.packages(c("ggplot2", "pheatmap","ggpubr"))
library("FactoMineR")
library("factoextra")

library(GSEABase)
library(GSVA)
library(clusterProfiler)
library(genefu)
library(ggplot2)
library(ggpubr)
library(hgu133plus2.db)
library(limma)
library(org.Hs.eg.db)
library(pheatmap)

第二讲：从geo下载数据得到表达谱矩阵 step1-download.R

rm(list = ls())  ## 魔幻操作，一键清空~
options(stringsAsFactors = F)#在调用as.data.frame的时，将stringsAsFactors设置为FALSE可以避免character类型自动转化为factor类型
# 注意查看下载文件的大小，检查数据 
f='GSE64634_eSet.Rdata'
# https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE64634
library(GEOquery)
# 这个包需要注意两个配置，一般来说自动化的配置是足够的。
#Setting options('download.file.method.GEOquery'='auto')
#Setting options('GEOquery.inmemory.gpl'=FALSE)
if(!file.exists(f)){
  gset <- getGEO('GSE64634', destdir=".",
                 AnnotGPL = F,     ## 注释文件
                 getGPL = F)       ## 平台文件
  save(gset,file=f)   ## 保存到本地
}
load('GSE64634_eSet.Rdata')  ## 载入数据
class(gset)  #查看数据类型
length(gset)  #
class(gset[[1]])
gset
# assayData: 54675 features, 16 samples 

# 因为这个GEO数据集只有一个GPL平台，所以下载到的是一个含有一个元素的list
a=gset[[1]] #
dat=exprs(a) #a现在是一个对象，取a这个对象通过看说明书知道要用exprs这个函数
dim(dat)#看一下dat这个矩阵的维度
# GPL13667
dat[1:4,1:4] #查看dat这个矩阵的1至4行和1至4列，逗号前为行，逗号后为列
boxplot(dat,las=2)
dat=dat[apply(dat,1,sd)>0,]
dat[dat<0]=1
boxplot(dat,las=2)

#dat=log2(dat+1)
#boxplot(dat,las=2)
library(limma)
dat=normalizeBetweenArrays(dat)
boxplot(dat,las=2)
pd=pData(a) 
#通过查看说明书知道取对象a里的临床信息用pData
## 挑选一些感兴趣的临床表型。
group_list=c(rep('normal',4),rep('npc',12))
table(group_list)

dat[1:4,1:4] 

# GPL570

if(F){
  library(GEOquery)
  #Download GPL file, put it in the current directory, and load it:
  gpl <- getGEO('GPL570', destdir=".")
  colnames(Table(gpl))  
  head(Table(gpl)[,c(1,15)]) ## you need to check this , which column do you need
  probe2gene=Table(gpl)[,c(1,15)]
  head(probe2gene)
  library(stringr)  
  save(probe2gene,file='probe2gene.Rdata')
}
# 
# load(file='probe2gene.Rdata')
# ids=probe2gene 

library(hgu133plus2.db)
ids=toTable(hgu133plus2SYMBOL) #toTable这个函数：通过看hgu133plus2.db这个包的说明书知道提取probe_id（探针名）和symbol（基因名）的对应关系的表达矩阵的函数为toTable
head(ids) #head为查看前六行

head(ids)
colnames(ids)=c('probe_id','symbol')  
ids=ids[ids$symbol != '',]
ids=ids[ids$probe_id %in%  rownames(dat),]

dat[1:4,1:4]   
dat=dat[ids$probe_id,] 

ids$median=apply(dat,1,median) #ids新建median这一列，列名为median，同时对dat这个矩阵按行操作，取每一行的中位数，将结果给到median这一列的每一行
ids=ids[order(ids$symbol,ids$median,decreasing = T),]#对ids$symbol按照ids$median中位数从大到小排列的顺序排序，将对应的行赋值为一个新的ids
ids=ids[!duplicated(ids$symbol),]#将symbol这一列取取出重复项，'!'为否，即取出不重复的项，去除重复的gene ，保留每个基因最大表达量结果s
dat=dat[ids$probe_id,] #新的ids取出probe_id这一列，将dat按照取出的这一列中的每一行组成一个新的dat
rownames(dat)=ids$symbol#把ids的symbol这一列中的每一行给dat作为dat的行名
dat[1:4,1:4]  #保留每个基因ID第一次出现的信息


save(dat,group_list,file = 'step1-output.Rdata')

step2-check.R

rm(list = ls())  ## 魔幻操作，一键清空~
options(stringsAsFactors = F)
load(file = 'step1-output.Rdata')
table(group_list)
# 每次都要检测数据
dat[1:4,1:4]
## 下面是画PCA的必须操作，需要看说明书。
dat=t(dat)#画PCA图时要求是行名时样本名，列名时探针名，因此此时需要转换
dat=as.data.frame(dat)#将matrix转换为data.frame
dat=cbind(dat,group_list) #cbind横向追加，即将分组信息追加到最后一列
library("FactoMineR")#画主成分分析图需要加载这两个包
library("factoextra") 
# The variable group_list (index = 54676) is removed
# before PCA analysis
dat.pca <- PCA(dat[,-ncol(dat)], graph = FALSE)#现在dat最后一列是group_list，需要重新赋值给一个dat.pca,这个矩阵是不含有分组信息的
fviz_pca_ind(dat.pca,
             geom.ind = "point", # show points only (nbut not "text")
             col.ind = dat$group_list, # color by groups
             # palette = c("#00AFBB", "#E7B800"),
             addEllipses = TRUE, # Concentration ellipses
             legend.title = "Groups"
)
ggsave('all_samples_PCA.png')

step3-DEG.R

rm(list = ls()) 
options(stringsAsFactors = F)
load(file = 'step1-output.Rdata')
# 每次都要检测数据
dat[1:4,1:4] 
table(group_list) #table函数，查看group_list中的分组个数
#通过为每个数据集绘制箱形图，比较数据集中的数据分布
boxplot(dat[1,]~group_list) #按照group_list分组画箱线图

bp=function(g){         #定义一个函数g，函数为{}里的内容
  library(ggpubr)
  df=data.frame(gene=g,stage=group_list)
  p <- ggboxplot(df, x = "stage", y = "gene",
                 color = "stage", palette = "jco",
                 add = "jitter")
  #  Add p-value
  p + stat_compare_means()
}
dev.off()
bp(dat[1,]) ## 调用上面定义好的函数，避免同样的绘图代码重复多次敲。
bp(dat[2,])
bp(dat[3,])
bp(dat[4,])
dim(dat)

library(limma)
design=model.matrix(~factor( group_list ))
fit=lmFit(dat,design)
fit=eBayes(fit)
## 上面是limma包用法的一种方式 
options(digits = 4) #设置全局的数字有效位数为4
#topTable(fit,coef=2,adjust='BH') 
topTable(fit,coef=2,adjust='BH') 
## 但是上面的用法做不到随心所欲的指定任意两组进行比较

design <- model.matrix(~0+factor(group_list))
colnames(design)=levels(factor(group_list))
head(design)
exprSet=dat
rownames(design)=colnames(exprSet)
design
contrast.matrix<-makeContrasts("npc-normal",
                               levels = design)
contrast.matrix ##这个矩阵声明，我们要把 Tumor 组跟 Normal 进行差异分析比较

deg = function(exprSet,design,contrast.matrix){
  ##step1
  fit <- lmFit(exprSet,design)
  ##step2
  fit2 <- contrasts.fit(fit, contrast.matrix) 
  ##这一步很重要，大家可以自行看看效果
  
  fit2 <- eBayes(fit2)  ## default no trend !!!
  ##eBayes() with trend=TRUE
  ##step3
  tempOutput = topTable(fit2, coef=1, n=Inf)
  nrDEG = na.omit(tempOutput) 
  #write.csv(nrDEG2,"limma_notrend.results.csv",quote = F)
  head(nrDEG)
  return(nrDEG)
}

deg = deg(exprSet,design,contrast.matrix)

head(deg)

save(deg,file = 'deg.Rdata')

load(file = 'deg.Rdata')
head(deg)
bp(dat[rownames(deg)[1],])
## for volcano 
if(T){
  nrDEG=deg
  head(nrDEG)
  attach(nrDEG)
  plot(logFC,-log10(P.Value))
  library(ggpubr)
  df=nrDEG
  df$v= -log10(P.Value) #df新增加一列'v',值为-log10(P.Value)
  ggscatter(df, x = "logFC", y = "v",size=0.5)
  
  df$g=ifelse(df$P.Value>0.01,'stable', #if 判断：如果这一基因的P.Value>0.01，则为stable基因
              ifelse( df$logFC >2,'up', #接上句else 否则：接下来开始判断那些P.Value<0.01的基因，再if 判断：如果logFC >1.5,则为up（上调）基因
                      ifelse( df$logFC < -2,'down','stable') )#接上句else 否则：接下来开始判断那些logFC <1.5 的基因，再if 判断：如果logFC <1.5，则为down（下调）基因，否则为stable基因
  )
  table(df$g)
  df$name=rownames(df)
  head(df)
  ggscatter(df, x = "logFC", y = "v",size=0.5,color = 'g')
  ggscatter(df, x = "logFC", y = "v", color = "g",size = 0.5,
            label = "name", repel = T,
            #label.select = rownames(df)[df$g != 'stable'] ,
            label.select = c('TTC9', 'AQP3', 'CXCL11','PTGS2'), #挑选一些基因在图中显示出来
            palette = c("#00AFBB", "#E7B800", "#FC4E07") )
  ggsave('volcano.png')
  
  ggscatter(df, x = "AveExpr", y = "logFC",size = 0.2)
  df$p_c = ifelse(df$P.Value<0.001,'p<0.001',
                  ifelse(df$P.Value<0.01,'0.001<p<0.01','p>0.01'))
  table(df$p_c )
  ggscatter(df,x = "AveExpr", y = "logFC", color = "p_c",size=0.2, 
            palette = c("green", "red", "black") )
  ggsave('MA.png')
  
  
}

## for heatmap 
if(T){ 
  load(file = 'step1-output.Rdata')
  # 每次都要检测数据
  dat[1:4,1:4]
  table(group_list)
  x=deg$logFC #deg取logFC这列并将其重新赋值给x
  names(x)=rownames(deg) #deg取probe_id这列，并将其作为名字给x
  cg=c(names(head(sort(x),100)),#对x进行从小到大排列，取前100及后100，并取其对应的探针名，作为向量赋值给cg
       names(tail(sort(x),100)))
  library(pheatmap)
  pheatmap(dat[cg,],show_colnames =F,show_rownames = F) #对dat按照cg取行，所得到的矩阵来画热图
  n=t(scale(t(dat[cg,])))#通过“scale”对log-ratio数值进行归一化，现在的dat是行名为探针，列名为样本名，由于scale这个函数应用在不同组数据间存在差异时，需要行名为样本，因此需要用t(dat[cg,])来转换，最后再转换回来
  
  n[n>2]=2
  n[n< -2]= -2
  n[1:4,1:4]
  pheatmap(n,show_colnames =F,show_rownames = F)
  ac=data.frame(g=group_list)
  rownames(ac)=colnames(n) #将ac的行名也就分组信息（是‘no TNBC’还是‘TNBC’）给到n的列名，即热图中位于上方的分组信息
  pheatmap(n,show_colnames =F,
           show_rownames = F,
          cluster_cols = F, 
           annotation_col=ac,filename = 'heatmap_top200_DEG.png') #列名注释信息为ac即分组信息
  
  
}

write.csv(deg,file = 'deg.csv')

step4-anno-go-kegg.R

rm(list = ls()) 
options(stringsAsFactors = F)

load(file = 'deg.Rdata')
head(deg)
## 不同的阈值，筛选到的差异基因数量就不一样，后面的超几何分布检验结果就大相径庭。
logFC_t=1.5
deg$g=ifelse(deg$P.Value>0.05,'stable',
            ifelse( deg$logFC > logFC_t,'UP',
                    ifelse( deg$logFC < -logFC_t,'DOWN','stable') )
)
table(deg$g)
head(deg)
deg$symbol=rownames(deg)
library(ggplot2)
library(clusterProfiler)
library(org.Hs.eg.db)
df <- bitr(unique(deg$symbol), fromType = "SYMBOL",
           toType = c( "ENTREZID"),
           OrgDb = org.Hs.eg.db)
head(df)
DEG=deg
head(DEG)

DEG=merge(DEG,df,by.y='SYMBOL',by.x='symbol')
head(DEG)
save(DEG,file = 'anno_DEG.Rdata')


gene_up= DEG[DEG$g == 'UP','ENTREZID'] 
gene_down=DEG[DEG$g == 'DOWN','ENTREZID'] 
gene_diff=c(gene_up,gene_down)
gene_all=as.character(DEG[ ,'ENTREZID'] )
data(geneList, package="DOSE")
head(geneList)
boxplot(geneList)
boxplot(DEG$logFC)

geneList=DEG$logFC
names(geneList)=DEG$ENTREZID
geneList=sort(geneList,decreasing = T)

source('kegg_and_go_up_and_down.R')
run_kegg(gene_up,gene_down,pro='npc_VS_normal')
# 需要多go数据库的3个条目进行3次富集分析，非常耗时。
# run_go(gene_up,gene_down,pro='npc_VS_normal')

go <- enrichGO(gene_up, OrgDb = "org.Hs.eg.db", ont="all") 
library(ggplot2)
library(stringr)
barplot(go, split="ONTOLOGY")+ facet_grid(ONTOLOGY~., scale="free") 
barplot(go, split="ONTOLOGY",font.size =10)+ 
  facet_grid(ONTOLOGY~., scale="free") + 
  scale_x_discrete(labels=function(x) str_wrap(x, width=50))+
  ggsave('gene_up_GO_all_barplot.png') 
go <- enrichGO(gene_down, OrgDb = "org.Hs.eg.db", ont="all") 
barplot(go, split="ONTOLOGY",font.size =10)+ 
  facet_grid(ONTOLOGY~., scale="free") + 
  scale_x_discrete(labels=function(x) str_wrap(x, width=50))+
  ggsave('gene_down_GO_all_barplot.png')

第三讲：对表达量矩阵用gsea软件做分析 step5-anno-GSEA.R

rm(list = ls()) 
options(stringsAsFactors = F)

library(ggplot2)
library(clusterProfiler)
library(org.Hs.eg.db)

### 对 MigDB中的全部基因集 做GSEA分析。
# http://www.bio-info-trainee.com/2105.html
# http://www.bio-info-trainee.com/2102.html 
{
  load(file = 'anno_DEG.Rdata')
  geneList=DEG$logFC
  names(geneList)=DEG$symbol
  geneList=sort(geneList,decreasing = T)
  #选择gmt文件（MigDB中的全部基因集）
  d='/home/mynewuser2/symbols/'
  gmts <- list.files(d,pattern = 'all')
  gmts
  #GSEA分析
  library(GSEABase) # BiocManager::install('GSEABase')
  ## 下面使用lapply循环读取每个gmt文件，并且进行GSEA分析
  ## 如果存在之前分析后保存的结果文件，就不需要重复进行GSEA分析。
  f='gsea_results.Rdata'
  if(!file.exists(f)){
    gsea_results <- lapply(gmts, function(gmtfile){
      # gmtfile=gmts[2]
      geneset <- read.gmt(file.path(d,gmtfile)) 
      print(paste0('Now process the ',gmtfile))
      egmt <- GSEA(geneList, TERM2GENE=geneset, verbose=FALSE)
      head(egmt)
      # gseaplot(egmt, geneSetID = rownames(egmt[1,]))
      
      return(egmt)
    })
    # 上面的代码耗时，所以保存结果到本地文件
    save(gsea_results,file = f)
  }
  load(file = f)
  #提取gsea结果，熟悉这个对象
  gsea_results_list<- lapply(gsea_results, function(x){
    cat(paste(dim(x@result)),'\n')
    x@result
  })
  while (!is.null(dev.list()))
  dev.off()
  gsea_results_df <- do.call(rbind, gsea_results_list)
  gseaplot(gsea_results[[2]],'KEGG_CELL_CYCLE') 
  gseaplot(gsea_results[[2]],'FARMER_BREAST_CANCER_CLUSTER_6') 
  gseaplot(gsea_results[[5]],'GO_CONDENSED_CHROMOSOME_OUTER_KINETOCHORE') 
  
}

step6-anno-GSVA.R

rm(list = ls()) 
options(stringsAsFactors = F)

library(ggplot2)
library(clusterProfiler)
library(org.Hs.eg.db)


### 对 MigDB中的全部基因集 做GSVA分析。
## 还有ssGSEA, PGSEA
{
  load(file = 'step1-output.Rdata')
  # 每次都要检测数据
  dat[1:4,1:4]  
  
  X=dat
  table(group_list)
  ## Molecular Signatures Database (MSigDb) 
  d='~/biosoft/MSigDB/symbols/'
  gmts=list.files(d,pattern = 'all')
  gmts
  library(ggplot2)
  library(clusterProfiler)
  library(org.Hs.eg.db)
  library(GSVA) # BiocManager::install('GSVA')
  
  if(T){
    es_max <- lapply(gmts, function(gmtfile){ 
      #gmtfile=gmts[8];gmtfile
      geneset <- getGmt(file.path(d,gmtfile))  
      es.max <- gsva(X, geneset, 
                     mx.diff=FALSE, verbose=FALSE, 
                     parallel.sz=1)
      return(es.max)
    })
    adjPvalueCutoff <- 0.001
    logFCcutoff <- log2(2)
    es_deg <- lapply(es_max, function(es.max){
      table(group_list)
      dim(es.max)
      design <- model.matrix(~0+factor(group_list))
      colnames(design)=levels(factor(group_list))
      rownames(design)=colnames(es.max)
      design
      library(limma)
      contrast.matrix<-makeContrasts(paste0(unique(group_list),collapse = "-"),
                                     levels = design)
      contrast.matrix<-makeContrasts("npc-normal",
                                     levels = design)
      
      contrast.matrix ##这个矩阵声明，我们要把progres.组跟stable进行差异分析比较
      
      deg = function(es.max,design,contrast.matrix){
        ##step1
        fit <- lmFit(es.max,design)
        ##step2
        fit2 <- contrasts.fit(fit, contrast.matrix) 
        ##这一步很重要，大家可以自行看看效果
        
        fit2 <- eBayes(fit2)  ## default no trend !!!
        ##eBayes() with trend=TRUE
        ##step3
        res <- decideTests(fit2, p.value=adjPvalueCutoff)
        summary(res)
        tempOutput = topTable(fit2, coef=1, n=Inf)
        nrDEG = na.omit(tempOutput) 
        #write.csv(nrDEG2,"limma_notrend.results.csv",quote = F)
        head(nrDEG)
        return(nrDEG)
      }
      
      re = deg(es.max,design,contrast.matrix)
      nrDEG=re
      head(nrDEG) 
      return(nrDEG)
    })
  } 
  
  gmts
  
  save(es_max,es_deg,file='gsva_msigdb.Rdata')
  
  load(file='gsva_msigdb.Rdata')
  
  library(pheatmap)
  lapply(1:length(es_deg), function(i){
    # i=2
    print(i)
    dat=es_max[[i]]
    df=es_deg[[i]]
    df=df[df$P.Value<0.01 & abs(df$logFC) > 0.3,]
    print(dim(df))
    if(nrow(df)>5){
      n=rownames(df)
      dat=dat[match(n,rownames(dat)),]
      ac=data.frame(g=group_list)
      rownames(ac)=colnames(dat)
      rownames(dat)=substring(rownames(dat),1,50)
      pheatmap::pheatmap(dat, 
                         fontsize_row = 8,height = 11,
                         annotation_col = ac,show_colnames = F,
                         filename = paste0('gsva_',strsplit(gmts[i],'[.]')[[1]][1],'.pdf'))
      
    }
  })
  
  adjPvalueCutoff <- 0.001
  logFCcutoff <- log2(2)
  df=do.call(rbind ,es_deg)
  es_matrix=do.call(rbind ,es_max)
  df=df[df$P.Value<0.01 & abs(df$logFC) > 0.5,]
  write.csv(df,file = 'GSVA_DEG.csv')
}

step7-visualization.R

rm(list = ls()) 
options(stringsAsFactors = F)

library(ggplot2)
library(clusterProfiler)
library(org.Hs.eg.db)

load(file = 'deg.Rdata')
head(deg)
load(file = 'anno_DEG.Rdata')

gene_up= DEG[DEG$g == 'UP','ENTREZID'] 
gene_down=DEG[DEG$g == 'DOWN','ENTREZID'] 
gene_diff=c(gene_up,gene_down)
gene_all=as.character(DEG[ ,'ENTREZID'] )
data(geneList, package="DOSE")
head(geneList)
boxplot(geneList)
boxplot(DEG$logFC)

geneList=DEG$logFC
names(geneList)=DEG$ENTREZID
geneList=sort(geneList,decreasing = T)

gene_down
gene_up
enrichKK <- enrichKEGG(gene         =  gene_up,
                      organism     = 'hsa',
                      #universe     = gene_all,
                      pvalueCutoff = 0.1,
                      qvalueCutoff =0.1)
head(enrichKK)[,1:6] 
browseKEGG(enrichKK, 'hsa04512')
dotplot(enrichKK)
enrichKK=DOSE::setReadable(enrichKK, OrgDb='org.Hs.eg.db',keyType='ENTREZID')
enrichKK 

#(3)可视化
#条带图
par(mfrow=c(2,1))
barplot(enrichKK,showCategory=20)
#气泡图
dotplot(enrichKK)
#下面的图需要映射颜色，设置和示例数据一样的geneList
geneList = deg$logFC
names(geneList)=deg$ENTREZID
geneList = sort(geneList,decreasing = T)
#(3)展示top5通路的共同基因，要放大看。
#Gene-Concept Network
cnetplot(enrichKK, categorySize="pvalue", foldChange=geneList,colorEdge = TRUE)
cnetplot(enrichKK, foldChange=geneList, circular = TRUE, colorEdge = TRUE)
#Enrichment Map
emapplot(enrichKK)
#(4)展示通路关系,仅仅是针对于GO数据库结果。
# goplot(enrichKK)
#(5)Heatmap-like functional classification
heatplot(enrichKK,foldChange = geneList)

第四讲：根据分组信息作差异分析

第五讲：对差异基因结果作go/kegg超几何分布检验富集分析

第六讲：指定基因分组boxplot指定基因list画热图