miRNA分析流程学习(二)/TCGAmiRNA数据三大R包整合差异分析再学习

rm(list = ls())
proj = "TCGA-HNSC"
load(paste0(proj,"_miRNA_count.Rdata"))

exp <- miRNA_count

load("hnsc_clinical.Rdata")
clinical = hnsc_clinical

# exp1 = exp[rowSums(exp)>0,]
# nrow(exp1)

exp = exp[apply(exp, 1, function(x) sum(x > 0) > 0.5*ncol(exp)), ]
nrow(exp)

# TCGA数据处理方式1
library(tinyarray) # 小洁老师的R包可以快速划分肿瘤/非肿瘤组织
Group = make_tcga_group(exp)
table(Group)
# Group
# normal  tumor 
#     44    525 

save(exp,Group,proj,clinical,file = paste0(proj,"_miRNApre.Rdata"))
setwd("..")

rm(list = ls())

proj = "TCGA-HNSC"
load(paste0(proj,"_miRNApre.Rdata"))
table(Group)

#deseq2----
library(DESeq2)
colData <- data.frame(row.names =colnames(exp), # 创建colData数据框
                      condition=Group)
#注意事项:如果多次运行，表达矩阵改动过的话，需要从工作目录下删除下面if括号里的文件
if(!file.exists(paste0(proj,"_dd.Rdata"))){
  dds <- DESeqDataSetFromMatrix(
  countData = exp,
  colData = colData,
  design = ~ condition)
  dds <- DESeq(dds)
  save(dds,file = paste0(proj,"_dd.Rdata"))
}  
##因为上述步骤中的DESeq是计算差异的函数，数据量大了之后容易出现限速；
##因此为了避免每次加载都要等待，小洁老师用了一个if函数
##如果存在了R.data文件就不需要再加载了
load(file = paste0(proj,"_dd.Rdata"))
class(dds)
res <- results(dds, contrast = c("condition",rev(levels(Group))))
#constrast
c("condition",rev(levels(Group)))
class(res)
DEG1 <- as.data.frame(res)
library(dplyr)
DEG1 <- arrange(DEG1,pvalue)
DEG1 = na.omit(DEG1)
head(DEG1)

#添加change列标记基因上调下调
logFC_t = 1
pvalue_t = 0.05

k1 = (DEG1$pvalue < pvalue_t)&(DEG1$log2FoldChange < -logFC_t);table(k1)
k2 = (DEG1$pvalue < pvalue_t)&(DEG1$log2FoldChange > logFC_t);table(k2)
DEG1$change = ifelse(k1,"DOWN",ifelse(k2,"UP","NOT"))
table(DEG1$change)
head(DEG1)

#edgeR----
library(edgeR)

# edgeR 中的一个数据结构，用于存储基因表达数据和相关的样本信息。
dge <- DGEList(counts=exp,group=Group)
dge$samples$lib.size <- colSums(dge$counts)
dge <- calcNormFactors(dge) 

design <- model.matrix(~Group)

dge <- estimateGLMCommonDisp(dge, design)
dge <- estimateGLMTrendedDisp(dge, design)
dge <- estimateGLMTagwiseDisp(dge, design)

fit <- glmFit(dge, design)
fit <- glmLRT(fit) 

DEG2=topTags(fit, n=Inf)
class(DEG2)
DEG2=as.data.frame(DEG2)
head(DEG2)

k1 = (DEG2$PValue < pvalue_t)&(DEG2$logFC < -logFC_t)
k2 = (DEG2$PValue < pvalue_t)&(DEG2$logFC > logFC_t)
DEG2$change = ifelse(k1,"DOWN",ifelse(k2,"UP","NOT"))

head(DEG2)
table(DEG2$change)

#limma----
library(limma)
dge <- edgeR::DGEList(counts=exp)
dge <- edgeR::calcNormFactors(dge)
design <- model.matrix(~Group)
v <- voom(dge,design, normalize="quantile")

fit <- lmFit(v, design)
fit= eBayes(fit)

DEG3 = topTable(fit, coef=2, n=Inf)
DEG3 = na.omit(DEG3)

k1 = (DEG3$P.Value < pvalue_t)&(DEG3$logFC < -logFC_t)
k2 = (DEG3$P.Value < pvalue_t)&(DEG3$logFC > logFC_t)
DEG3$change = ifelse(k1,"DOWN",ifelse(k2,"UP","NOT"))
table(DEG3$change)
head(DEG3)

tj = data.frame(deseq2 = as.integer(table(DEG1$change)),
           edgeR = as.integer(table(DEG2$change)),
           limma_voom = as.integer(table(DEG3$change)),
           row.names = c("down","not","up")
          );tj
save(DEG1,DEG2,DEG3,Group,tj,file = paste0(proj,"_DEG.Rdata"))

library(ggplot2)
library(tinyarray)
exp[1:4,1:4]
# cpm 去除文库大小的影响
dat = log2(cpm(exp)+1)
pca.plot = draw_pca(dat,Group, addEllipses = T);pca.plot ##加了addellipase参数取消了椭圆
save(pca.plot,file = paste0(proj,"_pcaplot.Rdata"))

cg1 = rownames(DEG1)[DEG1$change !="NOT"]
cg2 = rownames(DEG2)[DEG2$change !="NOT"]
cg3 = rownames(DEG3)[DEG3$change !="NOT"]

h1 = draw_heatmap(dat[cg1,],Group)
h2 = draw_heatmap(dat[cg2,],Group)
h3 = draw_heatmap(dat[cg3,],Group)

v1 = draw_volcano(DEG1,pkg = 1,logFC_cutoff = logFC_t)
v2 = draw_volcano(DEG2,pkg = 2,logFC_cutoff = logFC_t)
v3 = draw_volcano(DEG3,pkg = 3,logFC_cutoff = logFC_t)

library(patchwork)
(h1 + h2 + h3) / (v1 + v2 + v3) +plot_layout(guides = 'collect') &theme(legend.position = "none")

ggsave(paste0(proj,"_heat_vo.png"),width = 15,height = 10)

UP=function(df){
  rownames(df)[df$change=="UP"]
}
DOWN=function(df){
  rownames(df)[df$change=="DOWN"]
}

up = intersect(intersect(UP(DEG1),UP(DEG2)),UP(DEG3))
down = intersect(intersect(DOWN(DEG1),DOWN(DEG2)),DOWN(DEG3))
dat = log2(cpm(exp)+1)
hp = draw_heatmap(dat[c(up,down),],Group)
##导出上下调基因


#上调、下调基因分别画维恩图
up_genes = list(Deseq2 = UP(DEG1),
          edgeR = UP(DEG2),
          limma = UP(DEG3))

down_genes = list(Deseq2 = DOWN(DEG1),
          edgeR = DOWN(DEG2),
          limma = DOWN(DEG3))

up.plot <- draw_venn(up_genes,"UPgene")
down.plot <- draw_venn(down_genes,"DOWNgene")

#维恩图拼图，终于搞定

library(patchwork)
#up.plot + down.plot
# 拼图
pca.plot + hp+up.plot +down.plot+ plot_layout(guides = "collect")
ggsave(paste0(proj,"_heat_ve_pca.pdf"),width = 15,height = 10)