comparison Testtest/GCMS-test_analyze.R @ 0:40de28c7d3fb draft

Uploaded

0:40de28c7d3fb

# author: Pauline Ribeyre


#####################
# required packages #
#####################

library("metaMS") # provides "runGC" function


############
# analysis #
############

peakspectra_table <- function(GC_results, file_title) {
  # Saves the pseudospectra in 2 files (text and tabular formats).
  
  names <- c()
  rts <- c()
  rt.sds <- c()
  mzs_df <- list()
  for (ps in GC_results$PseudoSpectra) {
    
    name <- ps$Name
    rt <- ps$rt
    rt.sd <- ps$rt.sd
    
    names <- c(names, name)
    rts <- c(rts, rt)
    rt.sds <- c(rt.sds, rt.sd)
    
    spectrum <- data.frame(ps$pspectrum)
    mz <- c()
    maxo <- c()
    for (i in 1:nrow(spectrum)) {
      ion <- spectrum[i,]
      mz <- c(mz, ion$mz)
      maxo <- c(maxo, ion$maxo)
    }
    df <- data.frame(name, rt, rt.sd, mz, maxo)
    mzs_df[[length(mzs_df) + 1]] <- df
    
  }
  
  df = do.call(rbind, mzs_df)
  
  write.table(df,
              file = file_title,
              quote = FALSE,
              row.names = FALSE,
              sep = "\t")
  
}


my_runGC <- function(n, cdf_files, titles_to_test, settings_to_test) {
  # Runs the data analysis and records the results.
  #
  # Args:
  #   n: index of the current test, to select the corresponding title and settings set.
  #   cdf_files: list of the data files' names.
  #   titles_to_test:  list of titles (one for each settings set) (concatenation of the values taken by the varied parameters).
  #   settings_to_test: list of settings sets for runGC.
  
  library("metaMS")
  
  settings <- settings_to_test[n][[1]]
  title <- titles_to_test[n]
  
  print(title)
  
  if (!file.exists(paste0("Peak_tables/", title, ".tsv"))) {

    # run
    GC_results <- runGC(files = cdf_files, settings = settings, returnXset = TRUE, nSlaves = 20)

    # order the result table by retention time
    peak_table <- GC_results$PeakTable <- GC_results$PeakTable[order(GC_results$PeakTable[,"rt"]),]
    peak_table_values <- peak_table[,5:(5 + length(cdf_files) - 1)]
    peak_table$nb_zeros <- apply(peak_table_values, 1, function(x) sum(x == 0))
    zeros_per_l <- sum(peak_table$nb_zeros)/nrow(peak_table)

    # record the table in a file
    file_title <- paste0("Peak_tables/", title, ".tsv")
    write.table(peak_table, file = file_title, sep = "\t", row.names = FALSE) # /!\ title length -> cannot open connexion
    
    # record the RData
    file_title <- paste0("RDatas/", title, ".RData")
    # save(GC_results, settings, file = file_title)
    save(GC_results, file = file_title)
    
    # record the pseudospectra in files (.msp and .tsv)
    file_title <- paste0("Pseudospectra/", title, ".msp")
    write.msp(GC_results$PseudoSpectra, file = file_title, newFile = TRUE)
    file_title <- paste0("Pseudospectra/", title, ".tsv")
    peakspectra_table(GC_results, file_title)

    cat(paste(zeros_per_l, "zeros per line.\n\n"))

  } # end if
  
}


runGC_vary_parameters_parallel <- function(nb_cores, cdf_files, settings, vary) {
  # Calculates the number of sets of paramaters and runs the analysis on several cores.
  #
  # Args:
  #   nb_cores: maximum number of cores to use.
  #   cdf_files: list of the data files' names.
  #   settings: default settings for runGC().
  #   vary: list of parameters to vary and the values each parameter must take.
  
  # calculate the number of possibilities with the parameters' ranges
  nb_possibilites <- 1
  for (param in vary) {
    range_param <- param[3:length(param)]
    nb_possibilites <- nb_possibilites * length(range_param)
  }
  cat("Settings variations:", nb_possibilites, "combinations.\n")
  
  dir.create("RDatas", showWarnings = FALSE) #create the folder where the RDatas will be saved
  dir.create("Peak_tables", showWarnings = FALSE) #create the folder where the peak tables will be saved
  # dir.create("Ions_per_intensity", showWarnings = FALSE) #create the folder where the nb of ions per intensity range will be saved
  dir.create("Pseudospectra", showWarnings = FALSE) #create the folder where the pseudospectra will be saved

  time.start <- Sys.time() # start the timer

  # run the function on several cores
  if (length(titles_to_test) < nb_cores)
    nb_cores <- length(titles_to_test)
  cluster <- makeCluster(nb_cores)#, outfile = "")
  parLapplyLB(cluster, 1:length(titles_to_test), my_runGC, cdf_files, titles_to_test, settings_to_test)
  print(titles_to_test)
  stopCluster(cluster)

  time.end <- Sys.time() # stop the timer
  Tdiff <- difftime(time.end, time.start)
  print(Tdiff)
  
}


############
#   main   #
############

runGC_vary_parameters_parallel(nb_cores, cdf_files, settings, vary_list)