diff Testtest/GCMS-test.R @ 0:40de28c7d3fb draft

Uploaded

--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/Testtest/GCMS-test.R	Thu Nov 23 08:50:14 2017 -0500
@@ -0,0 +1,220 @@
+# author: Pauline Ribeyre
+
+
+#####################
+# required packages #
+#####################
+
+library(batch) # provides "parseCommandargv" function
+
+
+############
+#   init   #
+############
+
+# read the parameters
+argv_list <- parseCommandArgs(evaluate = FALSE) #, trailingOnly = TRUE)
+argv <- unlist(argv_list)
+
+# common environment to share variables between files
+env <- new.env()
+
+# timestamp made unique by adding a random element
+env$timestamp <- round(as.numeric(Sys.time()) + runif(1, 0, 100))
+
+# set working directory
+tmpwd <- paste0("/galaxy/data/", env$timestamp)
+# system("chmod -R 0777 .")
+# dir.create(tmpwd, mode = "0777")
+dir.create(tmpwd)
+setwd(tmpwd)
+
+# readme file
+readme_msg <- paste0("\"", env$timestamp, "\"",
+                     " is a temporary folder storing data for an instance of Galaxy tool \"GCMS-test\" launched on ",
+                    Sys.time(), ". ",
+                    "This folder should be deleted automatically at the end of the tool's execution. ",
+                    "If the instance ended or has been aborted and this folder still exists, it can be deleted manually.")
+write(readme_msg, file = "README.txt")
+
+################
+# source files #
+################
+
+source_local <- function(fname){
+  # Gets a file's full name in Galaxy's directories.
+  #
+  # Args:
+  #   fname: the name of the file.
+  #
+  # Returns:
+  #   The full name of the file.
+
+  argv <- commandArgs(trailingOnly = FALSE)
+  base_dir <- dirname(substring(argv[grep("--file=", argv)], 8))
+  return (paste(base_dir, fname, sep="/"))
+
+}
+
+env$source_settings <- source_local("GCMS-test_settings.R")
+env$source_analyze <- source_local("GCMS-test_analyze.R")
+env$source_output <- source_local("GCMS-test_output.R")
+env$source_spectrum <- source_local("GCMS-test_spectrum.R")
+
+env$source_file_names <- "file_names.txt"
+
+
+#############
+# functions #
+#############
+
+args_to_vary_list <- function() {
+  # Creates the list of settings variations using the arguments.
+  #
+  # Returns:
+  #   The list of settings variations.
+
+  param <- c("step", "steps", "mzdiff", "fwhm", "simthresh", "snthresh", "max",
+             "minclassfraction", "minclasssize", "rtdiff") #, "minfeat")
+
+  PeakPicking_param <- c("step", "steps", "mzdiff", "fwhm", "snthresh", "max")
+  betweenSamples_param <- c("min.class.fraction", "min.class.size", "rtdiff", "simthresh")
+
+  vary <- list()
+
+  for (p in param) {
+
+    min <- as.numeric(argv[[paste0(p, "_min")]])
+    max <- as.numeric(argv[[paste0(p, "_max")]])
+    step <- as.numeric(argv[[paste0(p, "_step")]])
+
+    if (p == "minclassfraction")
+      p <- "min.class.fraction"
+    else if (p == "minclasssize")
+      p <- "min.class.size"
+
+    if (p %in% PeakPicking_param)
+      name <- paste0("PeakPicking$", p)
+    else if (p %in% betweenSamples_param)
+      name <- paste0("betweenSamples.", p)
+    else
+      stop("\"", p, "\" is not a valid parameter to vary.")
+
+    # prevent seq error
+    if (min > max)
+      max <- min
+
+    range <- seq(min, max, step)
+    vary[[length(vary) + 1]] <- c(p, name, range)
+
+  }
+
+  return (vary)
+
+}
+
+
+read_ions_to_check <- function() {
+  # Reads the arguments concerning the ions to check.
+  
+  env$ions_name <- list()
+  env$ions_rt <- list()
+  env$ions_mzs <- list()
+  
+  env$nb_ions_to_check <- length(grep("ions_name_", names(argv)))
+  
+  if (env$nb_ions_to_check > 0) {
+    for (i in 1:env$nb_ions_to_check) {
+      env$ions_name[length(env$ions_name) + 1] <- argv[[paste0("ions_name_", i)]]
+      env$ions_rt[length(env$ions_rt) + 1] <- as.numeric(argv[[paste0("ions_rt_", i)]])
+      mzs <- argv[[paste0("ions_mzs_", i)]]
+      mzs <- as.numeric(strsplit(mzs, ",")[[1]])
+      env$ions_mzs[length(env$ions_mzs) + 1] <- list(mzs)
+    }
+  }
+  
+}
+
+
+############
+#  inputs  #
+############
+
+env$nb_cores <- as.numeric(argv[["nb_cores"]])
+cat("Nb cores:", env$nb_cores, "\n")
+
+# get data from zip file
+if (!is.null(argv[["zip_file"]])) {
+  env$cdf_files <- unzip(argv[["zip_file"]])
+} else {
+  stop("No files have been provided.")
+}
+
+# vary parameters
+env$vary_list <- args_to_vary_list()
+
+# count duplicate ions
+env$count_duplicates <- argv[["count_duplicates"]] == "true"
+if (env$count_duplicates) {
+  
+  # delta rt
+  env$duplicates_delta_rt <- as.numeric(argv[["duplicates_delta_rt"]])
+  
+  # delta mz
+  env$duplicates_delta_mz <- as.numeric(argv[["duplicates_delta_mz"]])
+  
+}
+
+# check the presence of ions
+env$nb_ions_to_check <- 0
+env$check_ions <- argv[["check_ions"]] == "true"
+if (env$check_ions) {
+  
+  read_ions_to_check()
+  
+  # delta rt
+  env$ions_delta_rt <- as.numeric(argv[["ions_delta_rt"]])
+  
+  # delta mz
+  env$ions_delta_mz <- as.numeric(argv[["ions_delta_mz"]])
+  
+}
+
+
+#############
+#  outputs  #
+#############
+
+env$summary_out <- argv[["summary_out"]]
+env$peakspectra_out <- argv[["peakspectra_out"]]
+env$intensity_graph_out <- argv[["intensity_graph_out"]]
+
+if (env$count_duplicates)
+  env$count_duplicates_out <- argv[["duplicates_out"]]
+
+env$compare_graph_out <- argv[["compare_graph_out"]]
+
+
+##########
+#  run   #
+##########
+
+cat("---------- Settings ----------\n")
+sys.source(env$source_settings, env)
+
+
+cat("---------- Analysis ----------\n")
+sys.source(env$source_analyze, env)
+
+
+cat("----------  Output  ----------\n")
+sys.source(env$source_output, env)
+
+
+##########
+#  end   #
+##########
+
+# delete temporary files
+setwd("..")
+unlink(tmpwd, recursive = TRUE)