R/mitoriboumi.R

In kendomaniac/CASC: rCASC reproducible Classification Analysis of Single Cell Sequencing Data

#' @title MitoRiboUmi
#' @description This function executes a ubuntu docker showing ribosomal mitochondrial protein genes fraction in each cell
#' @param group, a character string. Two options: sudo or docker, depending to which group the user belongs
#' @param scratch.folder, a character string indicating the path of the scratch folder
#' @param file, a character string indicating the path of the file, with file name and extension included
#' @param separator, separator used in count file, e.g. '\\t', ','
#' @param gtf.name, name for the gtf file tu be used.
#' @param bio.type, ENSEMBL biotype of interest, default for single-cell protein_coding
#' @param umiXgene, a integer defining how many UMI are required to call a gene as present. default: 3
#' @author Luca Alessandri, alessandri [dot] luca1991 [at] gmail [dot] com, University of Torino
#'
#' @return a pdf file called Ribo_mito.pdf
#' @examples
#'\dontrun{
#'    system("wget ftp://ftp.ensembl.org/pub/release-94/gtf/mus_musculus/Mus_musculus.GRCm38.94.gtf.gz")
#'    system("gzip -d Mus_musculus.GRCm38.94.gtf.gz")
#'    system("wget http://130.192.119.59/public/testSCumi_mm10.csv.zip")
#'    unzip("testSCumi_mm10.csv.zip")
#'    mitoRiboUmi(group="docker", file=paste(getwd(), "testSCumi_mm10.csv", sep="/"), 
#'                 scratch.folder="/data/scratch", separator=",", umiXgene=3, 
#'                 gtf.name="Mus_musculus.GRCm38.94.gtf", bio.type="protein_coding")

#'}
#' @export
mitoRiboUmi <- function(group=c("sudo","docker"), scratch.folder, file,separator, gtf.name, bio.type, umiXgene){


  data.folder=dirname(file)
positions=length(strsplit(basename(file),"\\.")[[1]])
matrixNameC=strsplit(basename(file),"\\.")[[1]]
matrixName=paste(matrixNameC[seq(1,positions-1)],collapse="")
format=strsplit(basename(basename(file)),"\\.")[[1]][positions]



  #running time 1
  ptm <- proc.time()
  #setting the data.folder as working folder
  if (!file.exists(data.folder)){
    cat(paste("\nIt seems that the ",data.folder, " folder does not exist\n"))
    return(2)
  }

  #storing the position of the home folder
  home <- getwd()
  setwd(data.folder)
  #initialize status
  system("echo 0 > ExitStatusFile 2>&1")

  #testing if docker is running
  test <- dockerTest()
  if(!test){
    cat("\nERROR: Docker seems not to be installed in your system\n")
    system("echo 10 > ExitStatusFile 2>&1")
    setwd(home)
    return(10)
  }



  #check  if scratch folder exist
  if (!file.exists(scratch.folder)){
    cat(paste("\nIt seems that the ",scratch.folder, " folder does not exist\n"))
    system("echo 3 > ExitStatusFile 2>&1")
    setwd(data.folder)
    return(3)
  }
  tmp.folder <- gsub(":","-",gsub(" ","-",date()))
  scrat_tmp.folder=file.path(scratch.folder, tmp.folder)
  writeLines(scrat_tmp.folder,paste(data.folder,"/tempFolderID", sep=""))
  cat("\ncreating a folder in scratch folder\n")
  dir.create(file.path(scrat_tmp.folder))
  #preprocess matrix and copying files


if(separator=="\t"){
separator="tab"
}
  
  
system(paste("cp -r ",data.folder,"/* ",scrat_tmp.folder,sep=""))


  #executing the docker job
 params <- paste("--cidfile ",data.folder,"/dockerID -v ",scrat_tmp.folder,":/scratch -v ", data.folder, ":/data -d docker.io/repbioinfo/ribomitoumi Rscript /home/main.R ",matrixName,"  ",format," ",separator," ",gtf.name," ",bio.type," ",umiXgene,sep="")

resultRun <- runDocker(group=group, params=params)

  #waiting for the end of the container work
  if(resultRun==0){
  #  system(paste("cp ", scrat_tmp.folder, "/* ", data.folder, sep=""))
  }
  #running time 2
  ptm <- proc.time() - ptm
  dir <- dir(data.folder)
  dir <- dir[grep("run.info",dir)]
  if(length(dir)>0){
    con <- file("run.info", "r")
    tmp.run <- readLines(con)
    close(con)
    tmp.run[length(tmp.run)+1] <- paste("user run time mins ",ptm[1]/60, sep="")
    tmp.run[length(tmp.run)+1] <- paste("system run time mins ",ptm[2]/60, sep="")
    tmp.run[length(tmp.run)+1] <- paste("elapsed run time mins ",ptm[3]/60, sep="")
    writeLines(tmp.run,"run.info")
  }else{
    tmp.run <- NULL
    tmp.run[1] <- paste("run time mins ",ptm[1]/60, sep="")
    tmp.run[length(tmp.run)+1] <- paste("system run time mins ",ptm[2]/60, sep="")
    tmp.run[length(tmp.run)+1] <- paste("elapsed run time mins ",ptm[3]/60, sep="")

    writeLines(tmp.run,"run.info")
  }

  #saving log and removing docker container
  container.id <- readLines(paste(data.folder,"/dockerID", sep=""), warn = FALSE)
  system(paste("docker logs ", substr(container.id,1,12), " &> ",data.folder,"/", substr(container.id,1,12),".log", sep=""))
  system(paste("docker rm ", container.id, sep=""))


  #Copy result folder
  cat("Copying Result Folder")
  system(paste("cp -r ",scrat_tmp.folder,"/* ",data.folder,"/",sep=""))
  #removing temporary folder
  cat("\n\nRemoving the temporary file ....\n")
  system(paste("rm -R ",scrat_tmp.folder))
  system("rm -fR out.info")
  system("rm -fR dockerID")
  system("rm  -fR tempFolderID")
  system(paste("cp ",paste(path.package(package="rCASC"),"containers.txt",sep="/")," ",data.folder, sep=""))
  setwd(home)
}