R/Utils_ParseGenedata.R

In michaelwitting/masstrixR: Annotation of MS based metabolomics data

#' This funtion parses a Genedata .gda file and returns three data frames containing the data and row and column annotations
#'
#' @param pathTofile Path to .gda file that shall be loaded
#' @return Returns a list with the three different dataframes, index 1 is the data, index 2 column annotation, index 3 row annotations
#' @examples
#' xxx
#' @export
readGdaFile <- function(pathToFile) {

  # create data frames for data
  rowAnnoDf <- data.frame(stringsAsFactors = FALSE)
  dataDf <- data.frame()
  clusterNames <- NULL

  # read file line by line
  con <- file(pathToFile, "r")

  while (TRUE) {
    line <- readLines(con, n = 1)

    # exit loop
    if (length(line) == 0) {
      break
    }

    # check for different line content
    if (grepl("^Name", line)) {

      # get content splitted
      content <- stringr::str_split(line, "\t")[[1]]

      # isolate sample names and add to ColumnDf
      SampleNames <- as.character(content[2:(length(content) - noOfRowAnno)])
      colAnnoDf <- data.frame(SampleNames = character(length(content) - noOfRowAnno - 1))
      colAnnoDf["SampleNames"] <- SampleNames

      # isolate row annotation names
      rowAnnotationNames <- as.character(content[(length(content) - noOfRowAnno + 1):length(content)])
    } else if (grepl("^# Row Annotations:", line)) {

      # split and get number of row annotations
      noOfRowAnno <- as.numeric(stringr::str_extract(stringr::str_split(line, ":")[[1]][2], "\\d+"))
      print(as.numeric(stringr::str_extract(stringr::str_split(line, ":")[[1]][2], "\\d+")))

      # print(line)
    } else if (grepl("\\[[A-Z]\\]", line)) {

      # get content splitted
      content <- stringr::str_split(line, "\t")[[1]]

      # get new annotation
      colAnnotationName <- content[[1]][1]
      colAnnotation <- content[2:(length(SampleNames) + 1)]

      # add to annotation data frame
      colAnnoDf[paste(colAnnotationName)] <- colAnnotation
    } else if (grepl("^(Peak|Cluster)", line)) {

      # get content splitted
      content <- stringr::str_split(line, "\t")[[1]]

      # get cluster names
      clusterNames <- c(clusterNames, content[1])

      # get peak values
      peakValues <- as.numeric(content[2:(length(content) - noOfRowAnno)])
      rowAnnotations <- content[(length(content) - noOfRowAnno + 1):length(content)]

      # add to rowAnnoDf and dataDf
      rowAnnoDf <- rbind.data.frame(rowAnnoDf,
        rowAnnotations,
        stringsAsFactors = FALSE
      )

      dataDf <- rbind.data.frame(
        dataDf,
        peakValues
      )
    }
  }

  # close connection to file
  close(con)

  # adjust column headers
  colnames(dataDf) <- SampleNames
  colnames(rowAnnoDf) <- rowAnnotationNames

  row.names(dataDf) <- clusterNames
  row.names(rowAnnoDf) <- clusterNames

  # adjust data type
  for (name in colnames(colAnnoDf)) {
    if (grepl("\\[N|n\\]", name)) {
      print(name)
      colAnnoDf[name] <- as.numeric(unlist(colAnnoDf[name]))
    } else if (grepl("\\[C|c\\]", name)) {
      colAnnoDf[name] <- as.factor(unlist(colAnnoDf[name]))
    }
  }

  # change the m/z column names
  names(rowAnnoDf)[names(rowAnnoDf) == "m/z"] <- "m.z"
  rowAnnoDf$m.z <- as.numeric(rowAnnoDf$m.z)


  return(list(dataDf, colAnnoDf, rowAnnoDf))
}