{"id":24946,"date":"2024-01-17T16:56:28","date_gmt":"2024-01-17T08:56:28","guid":{"rendered":"http:\/\/www.biocloudservice.com\/wordpress\/?p=24946"},"modified":"2024-01-17T16:56:30","modified_gmt":"2024-01-17T08:56:30","slug":"%e5%b7%ae%e5%bc%82%e8%a1%a8%e8%be%be%e6%a0%b7%e6%9c%ac%e5%b0%91%e6%97%a0%e9%87%8d%e5%a4%8d%ef%bc%9f%e5%bf%ab%e4%bd%bf%e7%94%a8edger%ef%bc%8c%e5%b0%8f%e6%9e%9c%e4%b8%89%e5%88%86%e9%92%9f%e6%95%99","status":"publish","type":"post","link":"http:\/\/www.biocloudservice.com\/wordpress\/?p=24946","title":{"rendered":"\u5dee\u5f02\u8868\u8fbe\u6837\u672c\u5c11\u65e0\u91cd\u590d\uff1f\u5feb\u4f7f\u7528edgeR\uff0c\u5c0f\u679c\u4e09\u5206\u949f\u6559\u4f60\u4e0a\u624b"},"content":{"rendered":"

\u5b9e\u9a8c\u6570\u636e\u83b7\u5f97\u5e76\u4e0d\u5bb9\u6613\uff0c\u800c\u4e14\u6210\u672c\u9ad8\u6602\uff0c\u6709\u65f6\u53ea\u662f\u60f3\u5927\u81f4\u770b\u4e00\u4e0b\u51e0\u79cd\u6761\u4ef6\u7684\u5dee\u5f02\u8868\u8fbe\uff0c\u5728\u6b63\u5f0f\u7684\u5b9e\u9a8c\u524d\u505a\u4e00\u4e2a\u9884\u5b9e\u9a8c\u5f88\u6709\u5fc5\u8981\u3002\u540c\u65f6\u51fa\u4e8e\u6210\u672c\u8003\u91cf\uff0c\u6bcf\u79cd\u6761\u4ef6\u4e0d\u4f1a\u505a\u91cd\u590d\u6837\u672c\u3002\u4f46\u662f\u65e0\u91cd\u590d\u7684\u6837\u672c\uff0c\u65e0\u6cd5\u4f7f\u7528DESeq2\u3002\u4f46\u662f\u6ca1\u5173\u7cfb\uff0c\u4eca\u5929\u5c0f\u679c\u4ecb\u7ecd\u4e00\u4e2a\u5dee\u5f02\u8868\u8fbe\u5229\u5668\u2014\u2014edgeR\u3002<\/p>\n

<\/p>\n

\u5305\u4e0b\u8f7d\u4e0e\u5b89\u88c5<\/h3>\n

\u6211\u4eec\u9700\u8981\u4e0b\u8f7d\u5e76\u5b89\u88c5\u4e24\u4e2aR\u5305\uff0c\u4e00\u4e2a\u662fedgeR\uff0c\u53e6\u4e00\u4e2a\u662fairway\u3002airway\u662f\u4e00\u4e2a\u63d0\u4f9b\u7528\u4e8e\u5206\u6790\u7684\u6570\u636e\u96c6\u3002<\/p>\n

if (!requireNamespace(“BiocManager”, quietly = TRUE))
\ninstall.packages(“BiocManager”)<\/p>\n

BiocManager::install(“edgeR”)
\nBiocManager::install(“airway”)<\/p>\n

library(edgeR) #\u52a0\u8f7dedgeR
\nlibrary(airway) ##\u52a0\u8f7dairway<\/p>\n

\u6784\u5efaGDElist\u5bf9\u8c61<\/h3>\n

DGEList\u662fedgeR\u5206\u6790\u5dee\u5f02\u57fa\u56e0\u4e2d\u6700\u6838\u5fc3\u7684\u5bf9\u8c61. \u5b83\u63d0\u4f9b\u4e24\u7c7b\u4fe1\u606f: \u8868\u8fbe\u91cf\u77e9\u9635\u548c\u5206\u7ec4\u4fe1\u606f.<\/p>\n

\u4e3a\u4e86\u65b9\u4fbf\u5927\u5bb6\u91cd\u590d,\u5c0f\u679c\u8fd9\u91cc\u7684\u6570\u636e\u6765\u81ea\u4e8eairway. \u5bf9\u4e8e\u60a8\u7684\u6570\u636e, \u53ef\u4ee5\u7528read.table\u7b49\u51fd\u6570\u8fdb\u884c\u5bfc\u5165.<\/p>\n

data(“airway”)<\/p>\n

> expr <- assay(airway)
\n> meta <- colData(airway)
\n> nrow(expr)
\n[1] 64102
\n> ncol(expr)
\n[1] 8<\/p>\n

\u6211\u4eec\u53ef\u4ee5\u770b\u5230exp\u6570\u636e\u670964102 \u4e2a\u57fa\u56e0\u548c8\u4e2a\u6837\u672c\u7684\u77e9\u9635<\/p>\n

\u5c0f\u679c\u4e3a\u4e86\u6f14\u793a\u65e0\u91cd\u590d\u6837\u672c\uff0c\u8fd9\u91cc\u5c31\u7528\u6570\u636e\u7684\u524d\u4e24\u5217\u6784\u5efaDGEList\u5bf9\u8c61\u3002<\/p>\n

> counts <- expr[,1:2]
\n> group <- 1:2
\n> dge <- DGEList(counts=counts, group = group)<\/p>\n

\u6570\u636e\u8fc7\u6ee4<\/h3>\n

\u518d\u4e00\u6b21\u8f6c\u5f55\u8fc7\u7a0b\u4e2d\u4e0d\u53ef\u80fd\u6240\u6709\u7684\u57fa\u56e0\u90fd\u6709\u8868\u8fbe\uff0c\u56e0\u6b64\u9700\u8981\u9884\u5148\u8fc7\u6ee4\u6389\u6ca1\u6709\u8868\u8fbe\u7684\u57fa\u56e0\u3002<\/p>\n

> keep <- rowSums(cpm(dge)>1) >= 1
\n> dge <- dge[keep, , keep.lib.sizes=FALSE]
\n> nrow(dge)
\n[1] 14756<\/p>\n

\u81f3\u5c11\u6709\u4e00\u4e2a\u8868\u8fbe\u7684\u57fa\u56e0(CPM > 1)\u3002\uff0c\u53bb\u6389\u5168\u90e8\u662f0\uff0c\u6ca1\u6709\u8868\u8fbe\u7684\u57fa\u56e0\u3002\u57fa\u56e0\u6570\u5c31\u4ece64102\u964d\u523014756\u4e86\u3002<\/p>\n

\u6807\u51c6\u5316<\/h3>\n

\u907f\u514d\u6587\u5e93\u5927\u5c0f\u4e0d\u540c\u5bfc\u81f4\u7684\u5206\u6790\u8bef\u5dee\u3002\u6211\u4eec\u9700\u8981\u5bf9\u6570\u636e\u8fdb\u884c\u6807\u51c6\u5316\u3002<\/p>\n

y <- calcNormFactors(y)<\/p>\n

calcNormFactors\u8fd9\u4e2a\u51fd\u6570\u4f7f\u7528\u7684TMM(trimmed mean of M-values)\u6765\u8fdb\u884c\u6807\u51c6\u5316\u3002<\/p>\n

\u65e0\u91cd\u590d\u5dee\u5f02\u8868\u8fbe\u5206\u6790<\/h3>\n

\u5904\u7406\u597d\u6570\u636e\u4e4b\u540e\u7ec8\u4e8e\u6765\u5230\u672c\u6587\u7684\u91cd\u5934\u620f\u4e86\u3002<\/p>\n

\u4f7f\u7528\u5404\u79cd\u5dee\u5f02\u8868\u8fbe\u5206\u6790\u5de5\u5177\u7684\u76ee\u7684\u90fd\u662f\u4e3a\u4e86\u8ba1\u7b97)\u6837\u672c\u7ec4\u95f4\u7684dispersion(\u79bb\u6563\u503c\uff09\uff0c\u6709\u4e86\u79bb\u6563\u503c\u4e4b\u540e\u624d\u80fd\u80fd\u7b97\u51fap\u503c\u3002<\/p>\n

\u5728\u65e0\u91cd\u590d\u6837\u672c\u7684\u6761\u4ef6\u4e0b\uff0cedgeR\u8ba1\u7b97\u79bb\u6563\u503c\u662f\u8fd9\u6837\u7684\u3002\u6211\u4eec\u53ef\u4ee5\u8bbe\u7f6e\u4e00\u4e2a\u503cbcv\uff0cedgeR\u6765\u8ba1\u7b97\u5b83\u7684\u5e73\u65b9\uff0c\u8fd9\u6837\u5c31\u5f97\u5230\u5355\u6837\u672c\u7684\u79bb\u6563\u503c\u4e86\u3002edgeR\u5efa\u8bae\uff0c\u9488\u5bf9\u4eba\u7c7b\u6570\u636e\uff0c\u5efa\u8bae\u8bbe\u5b9a\u4e3a0.4\u3002\u5bf9\u4e8e\u9057\u4f20\u4e0a\u76f8\u4f3c\uff0c\u5efa\u8bae\u8bbe\u5b9a\u4e3a0.1\u3002\u5bf9\u4e8e\u6280\u672f\u91cd\u590d\u7684\u5b9e\u9a8c\uff0c\u5efa\u8bae\u8bbe\u5b9a\u4e3a0.01\u3002<\/p>\n

\u5c0f\u679c\u8fd9\u91cc\u7528\u7684\u662f\u4eba\u7c7b\u6570\u636e\uff0c\u6240\u4ee5\u8bbe\u7f6e\u4e3a0.4<\/p>\n

dge_bcv <- dge
\nbcv <- 0.4 #
\net <- exactTest(dge_bcv, dispersion = bcv ^ 2)<\/p>\n

\u7528decideTestsDGE\u4e0a\u8c03\u7684\u57fa\u56e0\u548c\u4e0b\u8c03\u7684\u6709\u591a\u5c11. \u5e76\u8bbe\u7f6epvalue\u4e3a0.05<\/p>\n

gene1 <- decideTestsDGE(et, p.value = 0.05, lfc = 0)
\nsummary(gene1)<\/p>\n

<\/p>\n

\u5dee\u5f02\u57fa\u56e0\u600e\u4e48\u8fd9\u4e48\u5c11\uff0c\u53ef\u80fd\u4e24\u7ec4\u5dee\u5f02\u4e0d\u660e\u663e\uff0c\u5c0f\u679c\u8fd8\u662f\u4e0b\u8c03\u4e00\u4e0b\u53c2\u6570\u5427\uff0c\u6539\u62100.2\u53ef\u80fd\u597d\u4e00\u70b9\u3002<\/p>\n

bcv <- 0.2
\net2 <- exactTest(dge_bcv, dispersion = bcv ^ 2)
\ngene2 <- decideTestsDGE(et2, p.value = 0.05, lfc = 0)
\nsummary(gene2)<\/p>\n

<\/p>\n

\u8fd9\u65f6\u5019\u5dee\u5f02\u57fa\u56e0\u663e\u73b0\u51fa\u6765\u4e86~<\/p>\n

\u753b\u4e2a\u706b\u5c71\u56fe\u770b\u4e00\u4e0b<\/p>\n

library(ggplot2)
\ndf <- et2$table
\nggplot(df, aes(x=logFC, y=-log10(df$PValue)) ) +
\ngeom_point() +
\nylab(“-log10(p value)”) +
\ntheme_bw()<\/p>\n

<\/p>\n

\u548c\u6709\u91cd\u590d\u5dee\u5f02\u8868\u8fbe\u8fdb\u884c\u6bd4\u8f83<\/h3>\n

\u6211\u4eec\u6240\u4f7f\u7528\u7684airway\u6570\u636e\u96c6\u662f\u6709\u591a\u7ec4\u91cd\u590d\u7684\uff0c\u6240\u4ee5\u6211\u4eec\u53ef\u4ee5\u5f88\u65b9\u4fbf\u5730\u5bf9\u65e0\u91cd\u590d\u4e0e\u6709\u91cd\u590d\u8fdb\u884c\u6bd4\u8f83\uff0c\u770b\u4f1a\u76f8\u5dee\u591a\u5c11\u57fa\u56e0\u3002<\/p>\n

#\u4e0e\u6709\u91cd\u590d\u6bd4\u8f83
\ngroup <- meta$dex
\ny_rep <- DGEList(counts=expr, group = group)
\n#\u8fc7\u6ee4\u53ca\u6807\u51c6\u5316
\nkeep <- rowSums(cpm(y_rep)>1) >= 5
\ny_rep <- y_rep[keep, , keep.lib.sizes=FALSE]
\ny_rep <- calcNormFactors(y_rep)
\n#\u6784\u5efa\u8bbe\u8ba1\u77e9\u9635
\ndesign <- model.matrix(~group)
\ny_rep <- estimateDisp(y_rep, design = design)
\nfit <- glmQLFit(y_rep, design)
\nqlf.2vs1 <- glmQLFTest(fit, coef=2)
\n#\u67e5\u770b\u5dee\u5f02\u57fa\u56e0\u6570\u76ee
\nres <- decideTestsDGE(qlf.2vs1, p.value = 0.05, lfc = 0)
\nsummary(res)<\/p>\n

\u8fd9\u4e0b\u770b\u4e0b\u5dee\u5f02\u57fa\u56e0\u7684\u6570\u76ee\uff0cDown\u67091050\uff0cUp\u6709869\uff0c\u660e\u663e\u591a\u4e8e\u4ee5\u524d\u3002<\/p>\n

<\/p>\n

\u5c06\u6709\u91cd\u590d\u7684\u524d100\u5dee\u5f02\u57fa\u56e0\u548c\u65e0\u91cd\u590d\u7684\u524d100\u5dee\u5f02\u57fa\u56e0\u8fdb\u884c\u6bd4\u8f83<\/p>\n

c1 <- row.names(topTags(et2,n = 100 ))
\nc2 <- row.names(topTags(qlf.2vs1, n = 100 ))
\nintersect_gene <- intersect(c1,c2<\/p>\n

\u5728\u6709\u91cd\u590d\u5dee\u5f02\u8868\u8fbe\u7684\u706b\u5c71\u56fe\u4e0a\u9762\u6807\u8bb0\u4ea4\u96c6\u57fa\u56e0<\/p>\n

library(ggplot2)
\nlibrary(ggrepel)
\ndata <- qlf.2vs1$table
\ndata$significant <- as.factor(data$PValue<0.05 & abs(data$logFC) > 1)<\/p>\n

data2 <- data[intersect_gene,]
\ndata2$geneID <- rownames(data2)<\/p>\n

p <- ggplot(data=data, aes(x=logFC, y =-log10(PValue),color=significant)) +
\ngeom_point(alpha=0.8, size=1.2)+
\nscale_color_manual(values =c(“black”,”red”))+
\nlabs(title=”Volcanoplot”, x=”log2 (fold change)”,y=”-log10 (q-value)”)+
\ntheme(plot.title = element_text(hjust = 0.4))+
\ngeom_hline(yintercept = -log10(0.05),lty=4,lwd=0.6,alpha=0.8)+
\ngeom_vline(xintercept = c(1,-1),lty=4,lwd=0.6,alpha=0.8)+
\n#theme(legend.position=’none’)
\ntheme_bw()+
\ntheme(panel.border = element_blank(),
\npanel.grid.major = element_blank(),
\npanel.grid.minor = element_blank(),
\naxis.line = element_line(colour = “black”)) +
\ngeom_text(data=data2, aes(label=geneID),col=”red”,alpha = 1) +
\ngeom_text_repel(data=data2, aes(label=geneID),col=”black”,alpha = 0.8)<\/p>\n

print(p)<\/p>\n

\u8fd9\u6837\u5c31\u80fd\u6e05\u6670\u5730\u770b\u5230\u6211\u4eec\u65e0\u91cd\u590d\u7684\u57fa\u56e0\u4e0e\u6709\u91cd\u590d\u7684\u4ea4\u96c6\uff0c\u7684\u786e\u80fd\u627e\u5230\u6709\u660e\u663e\u5dee\u5f02\u8868\u8fbe\u7684\u57fa\u56e0\u3002\u4f46\u662f\u4e0d\u600e\u4e48\u660e\u663e\u7684\u4f46\u662f\u53c8\u6709\u5f71\u54cd\u7684\u57fa\u56e0\u5c31\u4f1a\u88ab\u7b5b\u9009\u6389\u3002<\/p>\n

<\/h3>\n

edgeR\u662f\u4e00\u79cd\u9002\u7528\u4e8e\u65e0\u91cd\u590d\u6837\u672c\u7684\u5dee\u5f02\u8868\u8fbe\u5206\u6790\u5de5\u5177\uff0c\u53ef\u4ee5\u7528\u6765\u5feb\u901f\u3001\u7b80\u5355\u5730\u8fdb\u884c\u9884\u5b9e\u9a8c\u3002\u4e0b\u9762\u662f\u4f7f\u7528edgeR\u8fdb\u884c\u5dee\u5f02\u8868\u8fbe\u5206\u6790\u7684\u6b65\u9aa4\uff1a<\/p>\n

    \n
  1. \u6570\u636e\u51c6\u5907\uff1a\u5c06\u539f\u59cb\u7684\u8868\u8fbe\u77e9\u9635\u5bfc\u5165\u5230R\u4e2d\uff0c\u5e76\u6839\u636e\u5b9e\u9a8c\u8bbe\u8ba1\u521b\u5efa\u4e00\u4e2a\u7ec4\u540d\u5411\u91cf\u3002<\/li>\n
  2. \u6570\u636e\u5904\u7406\uff1a\u4f7f\u7528edgeR\u7684\u51fd\u6570\u6765\u5bf9\u8868\u8fbe\u77e9\u9635\u8fdb\u884c\u6570\u636e\u5904\u7406\uff0c\u5305\u62ec\u6570\u636e\u6807\u51c6\u5316\u3001\u8fc7\u6ee4\u4f4e\u8868\u8fbe\u57fa\u56e0\u7b49\u3002\u53ef\u4ee5\u4f7f\u7528\u51fd\u6570like.EList\u3001calcNormFactors\u3001cpm\u7b49\u3002<\/li>\n
  3. \u5dee\u5f02\u8868\u8fbe\u5206\u6790\uff1a\u4f7f\u7528edgeR\u4e2d\u7684\u51fd\u6570glmQLFit\u548cglmQLFTest\u6765\u8fdb\u884c\u5dee\u5f02\u8868\u8fbe\u5206\u6790\uff0c\u5176\u4e2dglmQLFit\u7528\u4e8e\u62df\u5408\u5e7f\u4e49\u7ebf\u6027\u6a21\u578b\uff0cglmQLFTest\u7528\u4e8e\u8ba1\u7b97\u5dee\u5f02\u8868\u8fbe\u7684\u7edf\u8ba1\u663e\u8457\u6027\u3002<\/li>\n
  4. \u8c03\u6574bcv\u503c\uff1a\u4f7f\u7528edgeR\u4e2d\u7684\u51fd\u6570exactTest\u6765\u5bf9\u5dee\u5f02\u8868\u8fbe\u7ed3\u679c\u8fdb\u884c\u8c03\u6574\u3002<\/li>\n
  5. \u7ed3\u679c\u53ef\u89c6\u5316\uff1a\u4f7f\u7528edgeR\u4e2d\u7684\u51fd\u6570plotSmear\u3001plotMDS\u7b49\u6765\u8fdb\u884c\u5dee\u5f02\u8868\u8fbe\u7ed3\u679c\u7684\u53ef\u89c6\u5316\u3002<\/li>\n<\/ol>\n

    \u6700\u540e\u5c0f\u679c\u63d0\u9192\u4e00\u4e0b\uff0c\u7531\u4e8e\u65e0\u91cd\u590d\u6837\u672c\u7684\u4e0d\u786e\u5b9a\u6027\uff0c\u5dee\u5f02\u8868\u8fbe\u7ed3\u679c\u9700\u8981\u8c28\u614e\u89e3\u91ca\u3002\u867d\u7136edgeR\u53ef\u4ee5\u5728\u65e0\u91cd\u590d\u6837\u672c\u60c5\u51b5\u4e0b\u8fdb\u884c\u5dee\u5f02\u8868\u8fbe\u5206\u6790\uff0c\u4f46\u662f\u91cd\u590d\u6837\u672c\u4ecd\u7136\u662f\u66f4\u53ef\u9760\u7684\u9009\u62e9\uff0c\u56e0\u4e3a\u5b83\u53ef\u4ee5\u964d\u4f4e\u8bef\u5dee\uff0c\u63d0\u9ad8\u7ed3\u679c\u7684\u53ef\u9760\u6027\u3002<\/p>\n

    \u4f60\u662f\u5426\u5b66\u4f1a\u4e86\u5462\uff0c\u6b22\u8fce\u548c\u5c0f\u679c\u8ba8\u8bba~<\/p>\n

    \u5c0f\u679c\u7684\u4e91\u751f\u4fe1\u5e73\u53f0\u8fd8\u6709\u66f4\u591a\u65b9\u4fbf\u5b9e\u7528\u7684\u5de5\u5177\uff0c\u5feb\u6765\u77a7\u77a7\u5427<\/p>\n

    \u4e91\u751f\u4fe1\u5e73\u53f0\u94fe\u63a5\uff1ahttp:\/\/www.biocloudservice.com\/home.html<\/a><\/p>\n

    \u603b\u7ed3<\/h3>\n

    \u65e0\u91cd\u590d\u6837\u672c\u540c\u6837\u4e5f\u53ef\u4ee5\u7528edgeR\u8fdb\u884c\u5dee\u5f02\u8868\u8fbe\u5206\u6790\uff0c\u53ef\u4ee5\u7b80\u5355\u5730\u8fdb\u884c\u9884\u5b9e\u9a8c\u3002<\/p>\n