Add plotIdentGroup

workflow/scripts/createGroup.R

@@ -6,22 +6,16 @@ library(Seurat)
 library(readr)
 library(Matrix)
 
-#Get Seurat object input
+#Get param inputs
 option_list=list(
   make_option("--r",action="store",type='character',help="R data filename"),
   make_option("--o",action="store",type='character',help="Seurat object name"),
-  make_option("--g",action="store",type='character',help="Seurat group ident"),
-  make_option("--p",action="store",type='integer',default=0,help="Sample quantity")
+  make_option("--g",action="store",type='character',help="Seurat group ident")
 )
 opt=parse_args(OptionParser(option_list=option_list))
 rm(option_list)
 
-#Create output folders
-if (!dir.exists("./matrix")){
-  dir.create("./matrix")
-}
-
-#Load Seurat R data file and update
+#Load Seurat R data file and cells and update
 load(opt$r)
 data <- get(opt$o)
 rm(list=opt$o)
@@ -30,12 +24,10 @@ try(
     data <- UpdateSeuratObject(data)
   }
 )
+cells <- read.table("./cells.csv",quote="\"")
 
-#Downsample cells
-if (opt$p != 0){
-  Idents(data) <- "ALL"
-  data <- subset(data,cells=WhichCells(data,downsample=opt$p,seed=1))
-}
+#Filter cells
+data <- subset(data,cells=cells$V1)
 
 #Write ident group file
-write.table(data[[opt$g]],file=paste0("./matrix/group_",opt$g,".csv"),row.names=TRUE,col.names=FALSE,append=FALSE,sep=",")
\ No newline at end of file
+write.table(data[[opt$g]],file=paste0("./group_",opt$g,".csv"),row.names=TRUE,col.names=FALSE,append=FALSE,sep=",")
\ No newline at end of file

workflow/scripts/createMatrix.R

@@ -6,7 +6,7 @@ library(Seurat)
 library(readr)
 library(Matrix)
 
-#Get Seurat object input
+#Get param inputs
 option_list=list(
   make_option("--r",action="store",type='character',help="R data filename"),
   make_option("--o",action="store",type='character',help="Seurat object name"),
@@ -21,8 +21,11 @@ opt=parse_args(OptionParser(option_list=option_list))
 rm(option_list)
 
 #Create output folders
-if (!dir.exists("./matrix.raw")){
-  dir.create("./matrix.raw")
+if (!dir.exists("./matrix")){
+  dir.create("./matrix")
+}
+if (!dir.exists("./matrix/raw")){
+  dir.create("./matrix/raw")
 }
 
 #Load Seurat R data file and update
@@ -45,14 +48,15 @@ if (opt$p != 0){
 }
 
 #Write cell/gene matrix files
-writeMM(Matrix(GetAssayData(data,assay=opt$a,slot=opt$s),sparse=TRUE),file="./matrix.raw/expression.mtx")
-write.table(rownames(data),file=paste0("./matrix.raw/genes.csv"),row.names=FALSE,col.names=FALSE,append=FALSE,sep=",")
-write.table(colnames(data),file=paste0("./matrix.raw/cells.csv"),row.names=FALSE,col.names=FALSE,append=FALSE,sep=",")
+writeMM(Matrix(GetAssayData(data,assay=opt$a,slot=opt$s),sparse=TRUE),file="./matrix/raw/expression.mtx")
+write.table(rownames(data),file=paste0("./matrix/raw/genes.csv"),row.names=FALSE,col.names=FALSE,append=FALSE,sep=",")
+write.table(colnames(data),file=paste0("./matrix/raw/cells.csv"),row.names=FALSE,col.names=FALSE,append=FALSE,sep=",")
+write.table(colnames(data),file=paste0("./cells.csv"),row.names=FALSE,col.names=FALSE,append=FALSE,sep=",")
 
 #Write dimentionality reduction
 if (opt$t){
-  write.table(Embeddings(data,reduction="tsne"),file=paste0("./matrix.raw/tSNE.csv"),row.names=TRUE,col.names=NA,append=FALSE,sep=",")
+  write.table(Embeddings(data,reduction="tsne"),file=paste0("./matrix/raw/dr_tSNE.csv"),row.names=TRUE,col.names=NA,append=FALSE,sep=",")
 }
 if (opt$u){
-  write.table(Embeddings(data,reduction="umap"),file=paste0("./matrix.raw/UMAP.csv"),row.names=TRUE,col.names=NA,append=FALSE,sep=",")
+  write.table(Embeddings(data,reduction="umap"),file=paste0("./matrix/raw/dr_UMAP.csv"),row.names=TRUE,col.names=NA,append=FALSE,sep=",")
 }
\ No newline at end of file

workflow/scripts/plotIdentGroup.R

+if ("optparse" %in% row.names(installed.packages()) == FALSE){
+  install.packages("optparse",repos="http://cran.us.r-project.org")
+}
+library(optparse)
+library(readr)
+library(ggplot2)
+library(cowplot)
+
+#Get param inputs
+option_list=list(
+  make_option("--g",action="store",type='character',help="Seurat group ident"),
+  make_option("--t",action="store",type='logical',default=TRUE,help="tSNE calculated?"),
+  make_option("--u",action="store",type='logical',default=FALSE,help="UMAP calculated?")
+)
+opt=parse_args(OptionParser(option_list=option_list))
+rm(option_list)
+
+#Load group
+id <- read.csv(paste0("./group_",opt$g,".csv"),header=FALSE)
+names(id)[2] <- opt$g
+
+#Load dr
+if (opt$t == TRUE){
+  dr.tsne <- read.csv(paste0("./dr_tSNE.csv"),header=TRUE)
+  colnames(dr.tsne)[1] <- "Cells"
+  dr.tsne <- merge(x=dr.tsne,y=id,by.x="Cells",by.y="V1",all.x=TRUE)
+  png(paste0("ClusterGroup_tSNE_",opt$g,".png"),width=400,height=400,bg="white")
+  plot(ggplot(dr.tsne,aes_string(x="tSNE_1",y="tSNE_2",col=opt$g))+geom_point(size=2)+labs(col="ID")+theme_cowplot())
+  dev.off()
+}
+if (opt$u == TRUE){
+  dr.umap <- read.csv(paste0("./dr_UMAP.csv"),header=TRUE)
+  colnames(dr.umap)[1] <- "Cells"
+  dr.umap <- merge(x=dr.umap,y=id,by.x="Cells",by.y="V1",all.x=TRUE)
+  png(paste0("ClusterGroup_UMAP_",opt$g,".png"),width=400,height=400,bg="white")
+  plot(ggplot(dr.umap,aes_string(x="UMAP_1",y="UMAP_2",col=opt$g))+geom_point(size=2)+labs(col="ID")+theme_cowplot())
+  dev.off()
+}
\ No newline at end of file

workflow/scripts/renameGenes.R

@@ -5,20 +5,15 @@ library(optparse)
 library(readr)
 library(Matrix)
 
-#Get Seurat object input
+#Get param inputs
 option_list=list(
   make_option("--g",action="store",type='character',help="NCBI conversion filename")
 )
 opt=parse_args(OptionParser(option_list=option_list))
 rm(option_list)
 
-#Create output folders
-if (!dir.exists("./matrix")){
-  dir.create("./matrix")
-}
-
 #Load genes and conversion files
-genes <- read.table("./matrix.raw/genes.csv",quote="\"")
+genes <- read.table("./raw/genes.csv",quote="\"")
 convert <- read.delim(paste0("../../../workflow/files/",opt$g,".to.NCBI.txt"))
 
 #Filter gene to good conversions
@@ -32,18 +27,16 @@ genes.noambig.filtered <- genes.noambig[genes.noambig$V1 %in% genes$V1,]
 colnames(genes.noambig.filtered) <- c("GeneName","EntrezGene.ID")
 
 #Load matrix
-exp <- readMM("./matrix.raw/expression.mtx")
+exp <- readMM("./raw/expression.mtx")
 exp <- as.data.frame(as.matrix(exp))
 exp$gene <- genes$V1
 
 #Filter matrix
 exp.new <- merge(y=genes.noambig.filtered,x=exp,by.y="GeneName",by.x="gene",all.y=TRUE)
 exp.new <- as.data.frame(exp.new)
-#convert.final <- exp.new[,c(1,ncol(exp.new))]
 exp.new <- exp.new[,-1]
 
 #Write matrix and gene list
-write.table(as.character(exp.new$EntrezGene.ID),file="./matrix/genes.csv",append=FALSE,quote=TRUE,sep=",",row.names=FALSE,col.names=FALSE)
+write.table(as.character(exp.new$EntrezGene.ID),file="./genes.csv",append=FALSE,quote=TRUE,sep=",",row.names=FALSE,col.names=FALSE)
 exp.new.mm <- as(as.matrix(exp.new[,-ncol(exp.new)]),"dgTMatrix")
-writeMM(exp.new.mm,file="./matrix/expression.mtx")
-#write.table(convert.final,file="./analysis/matrix/convert.final.csv",append=FALSE,quote=TRUE,sep=",",row.names=FALSE,col.names=TRUE)
\ No newline at end of file
+writeMM(exp.new.mm,file="./expression.mtx")
\ No newline at end of file

workflow/scrna-seq_imgen.nf

@@ -15,24 +15,23 @@ params.convert = "GRCh38p5"
 params.outDir = "${baseDir}/../output"
 
 // Parse input variables
-rda = Channel
-    .fromPath(params.rda)
-    .ifEmpty { exit 1, "r data file not found: ${params.rda}" }
+rda = Channel.fromPath(params.rda)
 rda.into {
     rda_createMatrix
     rda_createGroup
 }
 seurat_createMatrix = params.seurat
 seurat_createGroup = params.seurat
-tsne = params.tsne
-umap = params.umap
+tsne_createMatrix = params.tsne
+tsne_plotIdentGroup = params.tsne
+umap_createMatrix = params.umap
+umap_plotIdentGroup = params.umap
 assay = params.assay
 slot = params.slot
 scale = params.scale
-sample_createMatrix = params.sample
-sample_createGroup = params.sample
-groups = file(params.groups)
-    .readLines()
+sample = params.sample
+groups_createGroup = file(params.groups).readLines()
+groups_plotIdentGroup = file(params.groups).readLines()
 convert = params.convert
 
 outDir = params.outDir
@@ -47,48 +46,47 @@ process createMatrix {
     input:
     file rda_createMatrix
     val seurat_createMatrix
-    val tsne
-    val umap
+    val tsne_createMatrix
+    val umap_createMatrix
     val assay
     val slot
     val scale
-    val sample_createMatrix
+    val sample
 
     output:
-    file("*") into matrixRaw
+    file("matrix/*") into matrixRaw
+    file("cells.csv") into cells
 
     script:
     """
     hostname
     ulimit -a
     module load seurat/3.0.0
-    seurat-Rscript ${baseDir}/scripts/createMatrix.R --r ${rda_createMatrix} --o ${seurat_createMatrix} --t ${tsne} --u ${umap} --a ${assay} --s ${slot} --l ${scale} --p ${sample_createMatrix}
+    seurat-Rscript ${baseDir}/scripts/createMatrix.R --r ${rda_createMatrix} --o ${seurat_createMatrix} --t ${tsne_createMatrix} --u ${umap_createMatrix} --a ${assay} --s ${slot} --l ${scale} --p ${sample}
     """
 }
 
-
 /*
 * createGroup: extract ident groups from Seurat object
 */
 process createGroup {
-    tag "${group}"
-    publishDir "${outDir}", mode: 'copy'
+    tag "${group_createGroup}"
+    publishDir "${outDir}/matrix", mode: 'copy', overwrite: true
 
     input:
     file rda_createGroup
-    val seurat_createGroup
-    each group from groups
-    val sample_createGroup
-
+    each group_createGroup from groups_createGroup
+    file cells
+    
     output:
-    file("*") into id
+    file("*") into id_plotIdentGroup
 
     script:
     """
     hostname
     ulimit -a
     module load seurat/3.0.0
-    seurat-Rscript ${baseDir}/scripts/createGroup.R --r ${rda_createGroup} --o ${seurat_createGroup} --g ${group} --p ${sample_createGroup}
+    seurat-Rscript ${baseDir}/scripts/createGroup.R --r ${rda_createGroup} --o ${seurat_createGroup} --g ${group_createGroup}
     """
 }
 
@@ -97,32 +95,61 @@ process createGroup {
 */
 process renameGenes {
     tag "NCBI"
-    publishDir "${outDir}", mode: 'copy'
+    publishDir "${outDir}/matrix", mode: 'copy', overwrite: false
 
     input:
-    file ("*") from matrixRaw.collect()
+    file matrixRaw
     val convert
 
     output:
-    file("*") into matrixFinal
+    file("*") into matrix_plotIdentGroup
+
 
     script:
     if (convert == "NO"){
         """
         hostname
         ulimit -a
-        mkdir matrix
-        cp -R ./matrix.raw/*.mtx ./matrix
-        cp -R ./matrix.raw/*.csv ./matrix
+        cp -R ./raw/*.mtx .
+        cp -R ./raw/*.csv .
         """
     } else {
         """
         hostname
         ulimit -a
-        mkdir matrix
-        cp -R ./matrix.raw/*.csv ./matrix
+        cp -R ./raw/*.csv .
         module load seurat/3.0.0
         seurat-Rscript ${baseDir}/scripts/renameGenes.R --g ${convert}
         """
     }
+}
+
+// Duplicate matrix files
+/*
+matrixFinal.into {
+    matrix_plotIdentGroup
+}
+*/
+/*
+* plotIdentGroup: plot identity groups in reduced dimensions
+*/
+process plotIdentGroup {
+    tag "${group_plotIdentGroup}"
+    publishDir "${outDir}/png/IdentGroup", mode: 'copy', overwrite: true
+
+    input:
+    each group_plotIdentGroup from groups_plotIdentGroup
+    file matrix_plotIdentGroup
+    file("*") from id_plotIdentGroup.collect()
+
+    output:
+    file("*.png") into plotsIdentGroup
+
+    script:
+    """
+    hostname
+    ulimit -a
+    module load seurat/3.0.0
+    seurat-Rscript ${baseDir}/scripts/plotIdentGroup.R --g ${group_plotIdentGroup}
+    """
 }
\ No newline at end of file