GSE study examples

zwbao · May 16, 2016 · e3849f3 · e3849f3
1 parent c048257
commit e3849f3
Show file tree

Hide file tree

Showing 33 changed files with 581 additions and 488 deletions.
diff --git a/.Rhistory b/.Rhistory
diff --git a/8-DEG/MA.png b/8-DEG/MA.png
diff --git a/8-DEG/BCV.png → 8-DEG/example_input_output/BCV.png b/8-DEG/BCV.png → 8-DEG/example_input_output/BCV.png
diff --git a/8-DEG/DEG_gseaInput.R → 8-DEG/example_input_output/DEG_gseaInput.R b/8-DEG/DEG_gseaInput.R → 8-DEG/example_input_output/DEG_gseaInput.R
diff --git a/8-DEG/DESeq2/DESeq2.pdf → 8-DEG/example_input_output/DESeq2/DESeq2.pdf b/8-DEG/DESeq2/DESeq2.pdf → 8-DEG/example_input_output/DESeq2/DESeq2.pdf
diff --git a/8-DEG/DESeq2/github-R-code.txt → ...ple_input_output/DESeq2/github-R-code.txt b/8-DEG/DESeq2/github-R-code.txt → ...ple_input_output/DESeq2/github-R-code.txt
diff --git a/8-DEG/Female.matrix → 8-DEG/example_input_output/Female.matrix b/8-DEG/Female.matrix → 8-DEG/example_input_output/Female.matrix
diff --git a/8-DEG/example_input_output/MA.png b/8-DEG/example_input_output/MA.png
diff --git a/8-DEG/MDS.png → 8-DEG/example_input_output/MDS.png b/8-DEG/MDS.png → 8-DEG/example_input_output/MDS.png
diff --git a/8-DEG/Male.matrix → 8-DEG/example_input_output/Male.matrix b/8-DEG/Male.matrix → 8-DEG/example_input_output/Male.matrix
diff --git a/8-DEG/P53.cls → 8-DEG/example_input_output/P53.cls b/8-DEG/P53.cls → 8-DEG/example_input_output/P53.cls
diff --git a/8-DEG/P53_hgu95av2.gct → 8-DEG/example_input_output/P53_hgu95av2.gct b/8-DEG/P53_hgu95av2.gct → 8-DEG/example_input_output/P53_hgu95av2.gct
diff --git a/8-DEG/RAWvsNORM.png → 8-DEG/example_input_output/RAWvsNORM.png b/8-DEG/RAWvsNORM.png → 8-DEG/example_input_output/RAWvsNORM.png
diff --git a/...OM/2014-paper-voom-DEG-RNA-seq-counts.pdf → ...OM/2014-paper-voom-DEG-RNA-seq-counts.pdf b/...OM/2014-paper-voom-DEG-RNA-seq-counts.pdf → ...OM/2014-paper-voom-DEG-RNA-seq-counts.pdf
diff --git a/8-DEG/baySeq/baySeq.pdf → 8-DEG/example_input_output/baySeq/baySeq.pdf b/8-DEG/baySeq/baySeq.pdf → 8-DEG/example_input_output/baySeq/baySeq.pdf
diff --git a/8-DEG/baySeq/paper-bayseq.pdf → ...mple_input_output/baySeq/paper-bayseq.pdf b/8-DEG/baySeq/paper-bayseq.pdf → ...mple_input_output/baySeq/paper-bayseq.pdf
diff --git a/8-DEG/deseq.results.csv → 8-DEG/example_input_output/deseq.results.csv b/8-DEG/deseq.results.csv → 8-DEG/example_input_output/deseq.results.csv
diff --git a/8-DEG/deseq2.results.csv → ...G/example_input_output/deseq2.results.csv b/8-DEG/deseq2.results.csv → ...G/example_input_output/deseq2.results.csv
diff --git a/8-DEG/edgeR/stanford_bios221.txt → ...e_input_output/edgeR/stanford_bios221.txt b/8-DEG/edgeR/stanford_bios221.txt → ...e_input_output/edgeR/stanford_bios221.txt
diff --git a/8-DEG/edger_classic.results.csv → ...le_input_output/edger_classic.results.csv b/8-DEG/edger_classic.results.csv → ...le_input_output/edger_classic.results.csv
diff --git a/8-DEG/edger_glm.results.csv → ...xample_input_output/edger_glm.results.csv b/8-DEG/edger_glm.results.csv → ...xample_input_output/edger_glm.results.csv
diff --git a/8-DEG/limma/limma_usersguide.pdf → ...e_input_output/limma/limma_usersguide.pdf b/8-DEG/limma/limma_usersguide.pdf → ...e_input_output/limma/limma_usersguide.pdf
diff --git a/8-DEG/limma_notrend.results.csv → ...le_input_output/limma_notrend.results.csv b/8-DEG/limma_notrend.results.csv → ...le_input_output/limma_notrend.results.csv
diff --git a/8-DEG/limma_trend.results.csv → ...mple_input_output/limma_trend.results.csv b/8-DEG/limma_trend.results.csv → ...mple_input_output/limma_trend.results.csv
diff --git a/8-DEG/pasillaGenes.count.cls → ...ample_input_output/pasillaGenes.count.cls b/8-DEG/pasillaGenes.count.cls → ...ample_input_output/pasillaGenes.count.cls
diff --git a/8-DEG/pasillaGenes.count.gct → ...ample_input_output/pasillaGenes.count.gct b/8-DEG/pasillaGenes.count.gct → ...ample_input_output/pasillaGenes.count.gct
diff --git a/8-DEG/qc-dispersions.png → ...G/example_input_output/qc-dispersions.png b/8-DEG/qc-dispersions.png → ...G/example_input_output/qc-dispersions.png
diff --git a/8-DEG/qc-heatmap-samples.png → ...ample_input_output/qc-heatmap-samples.png b/8-DEG/qc-heatmap-samples.png → ...ample_input_output/qc-heatmap-samples.png
diff --git a/8-DEG/example_input_output/qc_dispersions.png b/8-DEG/example_input_output/qc_dispersions.png
diff --git a/8-DEG/voom.results.csv → 8-DEG/example_input_output/voom.results.csv b/8-DEG/voom.results.csv → 8-DEG/example_input_output/voom.results.csv
diff --git a/8-DEG/server.R b/8-DEG/server.R
@@ -188,7 +188,7 @@ DEG_DESeq2 <- function(exprSet=exprSet,group_list=group_list){
 								colData = colData,
 								design = ~ group_list)
 	dds <- DESeq(dds)
-	png("qc-dispersions.png", 1000, 1000, pointsize=20)
+	png("qc_dispersions.png", 1000, 1000, pointsize=20)
 	plotDispEsts(dds, main="Dispersion plot")
 	dev.off()
 	res <- results(dds)
@@ -235,7 +235,7 @@ DEG_DESeq <- function(exprSet=exprSet,group_list=group_list){
 	de=newCountDataSet(exprSet,group_list)
 	de=estimateSizeFactors(de)
 	de=estimateDispersions(de)
-	png("qc-dispersions.png", 1000, 1000, pointsize=20)
+	png("qc_dispersions.png", 1000, 1000, pointsize=20)
 	DESeq::plotDispEsts(de, main="Dispersion plot")
 	dev.off()
 
@@ -337,13 +337,17 @@ shinyServer(function(input, output) {
 	})
 	output$matrix_control  <- renderDataTable({
 		example_data=get_example_data()
-		example_data$matrix_a		
+		tmp=example_data$matrix_a		
+		tmp=cbind(ID=rownames(tmp),tmp)
+		tmp
 	})
 	output$matrix_case  <- renderDataTable({
 		example_data=get_example_data()
-		example_data$matrix_b		
+		tmp=example_data$matrix_b	
+		tmp=cbind(ID=rownames(tmp),tmp)
+		tmp		
 	})
-		output$DEG_results  <- renderDataTable({
+	output$DEG_results  <- renderDataTable({
 		example_data=get_example_data()
 		DEG_method=tolower(input$method)
 		exprSet=example_data$exprSet
@@ -368,4 +372,48 @@ shinyServer(function(input, output) {
 		DEG_results=cbind(symbol=rownames(DEG_results),DEG_results)
 		DEG_results
 	})
+	output$MA_plot <-renderImage({
+	 list(src = "MA.png",
+         contentType = 'image/png',
+         width = 400,
+         height = 300,
+         alt = "This is alternate text")
+	})
+	output$MDS_plot <-renderImage({
+	list(src = "MDS.png",
+         contentType = 'image/png',
+         width = 400,
+         height = 300,
+         alt = "This is alternate text")
+	})
+	output$BCV_plot <-renderImage({
+	list(src = "BCV.png",
+         contentType = 'image/png',
+         width = 400,
+         height = 300,
+         alt = "This is alternate text")
+	})
+	output$volcanoplot <-renderImage({
+	list(src = "volcanoplot.png",
+         contentType = 'image/png',
+         width = 400,
+         height = 300,
+         alt = "This is alternate text")
+	})
+	output$qc_dispersions <-renderImage({
+	list(src = "qc_dispersions.png",
+         contentType = 'image/png',
+         width = 400,
+         height = 300,
+         alt = "This is alternate text")
+	})
+	output$RAWvsNORM <-renderImage({
+	list(src = "RAWvsNORM.png",
+         contentType = 'image/png',
+         width = 400,
+         height = 300,
+         alt = "This is alternate text")
+	})
+
+
 })
diff --git a/8-DEG/ui.R b/8-DEG/ui.R
@@ -62,9 +62,11 @@ shinyUI(fluidPage(
 			tabPanel("group 1",dataTableOutput("matrix_control")),
 			tabPanel("group 2",dataTableOutput("matrix_case")),
 			tabPanel("results for the DEG",dataTableOutput("DEG_results")),
-			tabPanel("quality control",imageOutput("qc_values_distribution"),
-										imageOutput("qc_sample_clustering"),
-										imageOutput("MA_plot"),
+			tabPanel("quality control",imageOutput("MA_plot"),
+										imageOutput("MDS_plot"),
+										imageOutput("BCV_plot"),
+										imageOutput("RAWvsNORM"),
+										imageOutput("qc_dispersions"),
 										imageOutput("volcanoplot")
 
 			)

diff --git a/9-microarray-examples/GSE43837.R b/9-microarray-examples/GSE43837.R
@@ -0,0 +1,43 @@
+## GSE43837
+#exprSet_url="ftp://ftp.ncbi.nlm.nih.gov/geo/series/GSE43nnn/GSE43837/matrix/GSE43837_series_matrix.txt.gz"
+#a=read.table(exprSet_url,sep = '\t',header = T,comment.char = "!")
+library(GEOquery)
+GSE43837 <- getGEO('GSE43837', destdir=".",getGPL = F)
+## geneNames/sampleNames/pData/exprs
+exprSet=exprs(GSE43837[[1]])
+
+library("annotate") 
+platformDB='u133x3p.db'
+library(platformDB, character.only=TRUE)
+probeset <- featureNames(GSE43837[[1]])
+#EGID <- as.numeric(lookUp(probeset, platformDB, "ENTREZID"))
+SYMBOL <-  lookUp(probeset, platformDB, "SYMBOL")
+a=cbind(SYMBOL,exprSet)
+## remove the duplicated probeset 
+rmDupID <-function(a=matrix(c(1,1:5,2,2:6,2,3:7),ncol=6)){
+  exprSet=a[,-1]
+  rowMeans=apply(exprSet,1,function(x) mean(as.numeric(x),na.rm=T))
+  a=a[order(rowMeans,decreasing=T),]
+  exprSet=a[!duplicated(a[,1]),]
+  #
+  exprSet=exprSet[!is.na(exprSet[,1]),]
+  rownames(exprSet)=exprSet[,1]
+  exprSet=exprSet[,-1]
+  return(exprSet)
+}
+exprSet=rmDupID(a)
+rn=rownames(exprSet)
+exprSet=apply(exprSet,2,as.numeric)
+rownames(exprSet)=rn
+exprSet=log(exprSet)
+
+pdata=pData(GSE43837[[1]])
+group_list=unlist(lapply(as.character(pdata$title),function(x) strsplit(x,"_")[[1]][2]))
+library(limma) 
+design=model.matrix(~factor(group_list))
+fit=lmFit(exprSet,design)
+fit=eBayes(fit)
+#options(digits = 4)
+DEG_limma=topTable(fit,coef=2,adjust='BH',n=Inf) 
+
+