R/on_plate_selection.r

In MACSQuantifyR: Fast treatment of MACSQuantify FACS data

#' @export on_plate_selection
#' @importFrom graphics axis locator plot points text title legend par
#' @importFrom grDevices dev.copy dev.off png
#' @importFrom utils write.table
#' @importFrom stats median sd
#' @importFrom grDevices heat.colors
#' @importFrom methods new
#' @importFrom gridExtra grid.arrange
#'
#' @title Sort experimental design on graphical 96-well-plate
#' @description
#' Loads user's data, prompt a graphical representation of a 96
#' well plate and let the user select where the duplicates of each
#' condition were placed. Use for reordering excel file.
#' Plate image with selection can also be saved in the form of a png
#' file in the "specified_path/output_MQ" directory.

#' @name on_plate_selection
#' @rdname on_plate_selection
#' @aliases on_plate_selection
#' @usage
#' on_plate_selection(MACSQuant,number_of_replicates,number_of_conditions,
#' control=FALSE,save.files=FALSE)
#' @param MACSQuant object of class MACSQuant resulting of the function
#' load_maxQuant().
#' Contains the original data table
#' @param number_of_replicates For each condition, the number of duplicates
#' (must be the same for all conditions)
#' @param number_of_conditions The number of conditions tested
#' (eg: Drug 1 alone, Drug 2 alone)
#' @param control Is there a control in this experiment (eg: Staurosporin)
#' @param save.files Used to save the image in the output folder
#' @examples
#' print("run manually, requires user input")
#' # on_plate_selection(MACSQuant,3,5)
#' # let you select 5 conditions of 3 replicates each
#' @return A formatted report file along with intermediates results

#source("./R/function_toolbox.R")
# source("./R/barplot_data.R")


on_plate_selection <- function(MACSQuant,
                                number_of_replicates,
                                number_of_conditions,
                                control = FALSE,
                                save.files = FALSE) {

    # check if data is imported or not
    if (is.null(MACSQuant@my_data)) {
        stop(paste("Data missing: Please run load_MACSQuant or a",
            "new_class_MQ to create MACSQuant object", sep = " "))
    }

    # checking if arguments are correct
    if (!is.numeric(number_of_conditions) | !is.numeric(number_of_replicates)) {
        stop(paste("Both number_of_conditions and number_of_replicates",
            "must be numeric",
            sep = " "
        ))
    } else if (number_of_conditions <= 0 | number_of_replicates <= 0) {
        stop(paste("Both number_of_conditions and number_of_replicates",
            "must be positive",
            sep = " "
        ))
    } else {
        MACSQuant@param.experiment$number_of_replicates <- number_of_replicates
        MACSQuant@param.experiment$number_of_conditions <- number_of_conditions
    }

    # checking if plot options (xlabs) are present
    # and correctly formatted
    if (length(MACSQuant@param.experiment$c_names) == 0) {
        col <- rep(2, number_of_conditions)
    } else if (length(MACSQuant@param.experiment$c_names) ==
        number_of_conditions) {
        col <- heat.colors(length(MACSQuant@param.experiment$c_names))
    } else if (length(MACSQuant@param.experiment$c_names) !=
        number_of_conditions) {
        stop(paste("If you give a name to your conditions:",
            "all conditions except controls must be named",
            "(length(c_names)==number_of_conditions)\n",
            "else rm(c_names) to remove condition names",
            sep = " "
        ))
    }

    # starting messages
    message("...To quit press ESC...")
    message(paste("...You can now select your conditions replicates",
        "(without control condition replicates)...",
        sep = " "
    ))

    # GUI plate plot init
    well_letter <- init_GUI()

    # initialization of dataframes sorted matrix of replicates
    # and statistics
    sorted_matrix <- matrix(
        data = NA,
        ncol = number_of_replicates,
        nrow = number_of_conditions
    )
    statistics <- data.frame(
        Full.path.first = NA,
        WID.first = NA,
        Fluo.percent.plus = NA,
        Fluo.percent.minus = NA,
        sd_percent = NA,
        Cell.count.minus = NA,
        sd_count = NA, stringsAsFactors = FALSE
    )

    # converting xy plot selecction to well name
    message(paste("--> ", number_of_conditions, " conditions: ...", sep = ""))
    for (i in seq_len(number_of_conditions)) {
        # every condition compute statistics on replicates
        # locate replicates for condition i
        loc <- locator(type = "n", n = number_of_replicates)
        # ensure proper data selection within graph
        if (length(loc) == 2 & sum(loc$y < 8.5) == 3 & sum(loc$y > 0.5) == 3
        & sum(loc$y < 12.5) == 3 & sum(loc$x > 0.5) == 3) {
            sorted_matrix[i, ] <- to_well_names(loc, col[i], well_letter)
            if (length(unique(sorted_matrix[i,])) != number_of_replicates)
            {stop("Ambiguous selection, please run the function again")}
            matched <- match_id_line(MACSQuant, sorted_matrix[i, ])
            if (length(matched) != number_of_replicates * 2) {
                stop(c(
                    paste(
                        "Your file may not contains 2 gates",
                        "for one or more replicates in: ",
                        sep = " "
                    ),
                    paste(sorted_matrix[i, ], collapse = ","), "\nPlease",
                    "be sure to only have P1 and P1/P2 gates in your file"
                ))
            }
            statistics[i, ] <- compute_statistics(MACSQuant,
                matched,
                stats = "mean"
            )
            # print(statistics[i, ])
            message(paste(i, "...", sep = ""))
        } else {
            warning("Outside of selection field, select your data again")
            break
        }
    }
    message("OK\n")
    sorted_matrix_final <- sorted_matrix
    if (i != number_of_conditions) {
        stop("Process interrupted please start again")
    }
    message("\n--> Done: replicates identified")
    message("--> Done: statistics on each condition replicates")


    if (control == TRUE) {
        MACSQuant@param.experiment$control <- TRUE
        message("...You can now select your control replicates...")
        message("--> 1 control: ...")

        # check with users
        sorted_matrix_ctrl <- matrix(
            data = NA, ncol = number_of_replicates,
            nrow = 1
        )

        loc2 <- locator(type = "n", n = number_of_replicates)
        if (length(loc2) == 2) {
            sorted_matrix_ctrl[1, ] <- to_well_names(loc2, 1, well_letter)
            matched <- match_id_line(MACSQuant, sorted_matrix_ctrl[1, ])
            if (length(matched) != number_of_replicates * 2) {
                warning(c(
                    paste(
                        "Your file may not contains 2 gates",
                        "for one or more replicates in: ",
                        sep = " "
                    ),
                    paste(sorted_matrix_ctrl[1, ], collapse = ","),
                    "\nPlease",
                    "be sure to only have P1 and P1/P2 gates in your file"
                ))
            }
            statistics[i + 1, ] <- compute_statistics(MACSQuant, matched,
                stats = "mean"
            )
            message("OK...")

            # print(paste('selection',paste(sorted_matrix_ctrl[1,],collapse =
            # ','),sep=' '))
            message("\n--> Done: statistics on each control replicates")
            sorted_matrix_final <- rbind(sorted_matrix_final,
                sorted_matrix_ctrl)
            message(sorted_matrix_final)
        } else {
            stop("Outside of selection field, select your data again")

        }
    }
    MACSQuant@my_replicates_sorted <- sorted_matrix_final
    if (control == FALSE &
        (length(MACSQuant@param.experiment$c_names) == 0)) {
        legend("topright",
            inset = c(-0.3, 0.05), legend = c("Conditions"),
            pch = c(1), col = col[1])
        MACSQuant@param.output$plt.labels <- c(rep("Conditions",
            number_of_conditions))
        plt.labels <- c(rep("Conditions", number_of_conditions))
        statistics[, seq(3, 7)] <- vapply(statistics[, seq(3, 7)], function(x) {
            as.numeric(x)
        }, numeric(number_of_conditions))
    } else if (control == TRUE &
        (length(MACSQuant@param.experiment$c_names) == 0)) {
        legend("topright", inset = c(-0.3, 0.05), legend = c(
            "Conditions",
            "Control"
        ), pch = c(1), col = c(col[1], 1))
        MACSQuant@param.output$plt.labels <- c(
            rep("Conditions",
                number_of_conditions), "Control")
        plt.labels <- c(rep("Conditions", number_of_conditions), "Control")
        statistics[, seq(3, 7)] <- vapply(statistics[, seq(3, 7)], function(x) {
            as.numeric(x)
        }, numeric(number_of_conditions + 1))
    } else if (control == FALSE &
        (length(MACSQuant@param.experiment$c_names) != 0)) {
        legend("topright",
            inset = c(-0.3, 0.05), legend = MACSQuant@param.experiment$c_names,
            pch = c(1), col = col
        )
        MACSQuant@param.output$plt.labels <-
            rep(MACSQuant@param.experiment$c_names)
        plt.labels <- rep(MACSQuant@param.experiment$c_names)
        statistics[, seq(3, 7)] <- vapply(statistics[, seq(3, 7)], function(x) {
            as.numeric(x)
        }, numeric(number_of_conditions))
    } else if (control == TRUE &
        (length(MACSQuant@param.experiment$c_names) != 0)) {
        legend("topright", inset = c(-0.3, 0.05), legend = c(
            MACSQuant@param.experiment$c_names,
            "Control"
        ), pch = c(1), col = c(col, 1))
        MACSQuant@param.output$plt.labels <-
            c(MACSQuant@param.experiment$c_names, "Control")
        plt.labels <- c(MACSQuant@param.experiment$c_names, "Control")
        statistics[, seq(3, 7)] <- vapply(statistics[, seq(3, 7)], function(x) {
            as.numeric(x)
        }, numeric(number_of_conditions + 1))
    }


    if (save.files == TRUE) {
        MACSQuant@param.output$save.files <- TRUE
        MACSQuant <- create_output_folder(MACSQuant)
        dev.copy(png, paste(MACSQuant@param.output$path,
            "/outputMQ/plate_template.png", sep = ""),
        width = 600, height = 600,
        units = "px"
        )
        dev.off()
        message("--> Done: image saved")
        write.table(statistics, paste(MACSQuant@param.output$path,
            "/outputMQ/statistics.txt ", sep = ""),
        sep = "\t")
        message("--> Done: statistics table saved")
    }


    message("--> Done: replicates stored in variable my_replicates_sorted")
    MACSQuant@statistics <- statistics

    indices <- order_data(MACSQuant)
    MACSQuant@my_data_sorted <- MACSQuant@my_data[indices, ]

    if (MACSQuant@param.experiment$control == TRUE &
        length(MACSQuant@param.experiment$doses) != 0) {
        col <- c(col, 1)
    }  else if (MACSQuant@param.experiment$control == TRUE &
        length(MACSQuant@param.experiment$doses) == 0) {
        MACSQuant@param.experiment$doses <- seq_len(number_of_conditions + 1)
        col <- c(col, 1)
    } else if (MACSQuant@param.experiment$control == FALSE &
        length(MACSQuant@param.experiment$doses) == 0) {
        MACSQuant@param.experiment$doses <- seq_len(number_of_conditions)
    }

    MACSQuant@param.output$plt.col <- col
    if (length(MACSQuant@param.experiment$doses.alt) == 0) {
        for (flav in c("counts", "percent"))
        {
            p <- barplot_data(MACSQuant, col, plt.flavour = flav,
                plt.labels = plt.labels, plt.combo = FALSE)
            grid.arrange(p)
            if (save.files == TRUE) {
                ggsave(paste(MACSQuant@param.output$path,
                    "/outputMQ/barplot_", flav, "_pipeline.png", sep = ""),
                width = 15.875, height = 15.875,
                units = "cm", p)
                message("--> Done: image saved")
            }
        }

    } else {
        message("...You can now run barplot_data(...plt.combo = TRUE)...")
    }
    return(MACSQuant)
}




on_plate_selection.internal <- function(MACSQuant,
                               number_of_replicates,
                               number_of_conditions,
                               control = FALSE,
                               save.files = FALSE) {

  # check if data is imported or not
  if (is.null(MACSQuant@my_data)) {
    stop(paste("Data missing: Please run load_MACSQuant or a",
               "new_class_MQ to create MACSQuant object", sep = " "))
  }

  # checking if arguments are correct
  if (!is.numeric(number_of_conditions) | !is.numeric(number_of_replicates)) {
    stop(paste("Both number_of_conditions and number_of_replicates",
               "must be numeric",
               sep = " "
    ))
  } else if (number_of_conditions <= 0 | number_of_replicates <= 0) {
    stop(paste("Both number_of_conditions and number_of_replicates",
               "must be positive",
               sep = " "
    ))
  } else {
    MACSQuant@param.experiment$number_of_replicates <- number_of_replicates
    MACSQuant@param.experiment$number_of_conditions <- number_of_conditions
  }

  # checking if plot options (xlabs) are present
  # and correctly formatted
  if (length(MACSQuant@param.experiment$c_names) == 0) {
    col <- rep(2, number_of_conditions)
  } else if (length(MACSQuant@param.experiment$c_names) ==
             number_of_conditions) {
    col <- heat.colors(length(MACSQuant@param.experiment$c_names))
  } else if (length(MACSQuant@param.experiment$c_names) !=
             number_of_conditions) {
    stop(paste("If you give a name to your conditions:",
               "all conditions except controls must be named",
               "(length(c_names)==number_of_conditions)\n",
               "else rm(c_names) to remove condition names",
               sep = " "
    ))
  }

  # starting messages
  message("...To quit press ESC...")
  message(paste("...You can now select your conditions replicates",
                "(without control condition replicates)...",
                sep = " "
  ))

  # GUI plate plot init
  well_letter <- init_GUI()

  # initialization of dataframes sorted matrix of replicates
  # and statistics
  sorted_matrix <- matrix(
    data = NA,
    ncol = number_of_replicates,
    nrow = number_of_conditions
  )
  statistics <- data.frame(
    Full.path.first = NA,
    WID.first = NA,
    Fluo.percent.plus = NA,
    Fluo.percent.minus = NA,
    sd_percent = NA,
    Cell.count.minus = NA,
    sd_count = NA, stringsAsFactors = FALSE
  )

  # converting xy plot selecction to well name
  message(paste("--> ", number_of_conditions, " conditions: ...", sep = ""))
  for (i in seq_len(number_of_conditions)) {
    # every condition compute statistics on replicates
    # locate replicates for condition i
    loc <- locator(type = "n", n = number_of_replicates)
    # ensure proper data selection within graph
    if (length(loc) == 2 & sum(loc$y < 8.5) == 3 & sum(loc$y > 0.5) == 3
        & sum(loc$y < 12.5) == 3 & sum(loc$x > 0.5) == 3) {
      sorted_matrix[i, ] <- to_well_names(loc, col[i], well_letter)
      if (length(unique(sorted_matrix[i,])) != number_of_replicates)
      {stop("Ambiguous selection, please run the function again")}
      matched <- match_id_line(MACSQuant, sorted_matrix[i, ])
      if (length(matched) != number_of_replicates * 2) {
        stop(c(
          paste(
            "Your file may not contains 2 gates",
            "for one or more replicates in: ",
            sep = " "
          ),
          paste(sorted_matrix[i, ], collapse = ","), "\nPlease",
          "be sure to only have P1 and P1/P2 gates in your file"
        ))
      }
      statistics[i, ] <- compute_statistics(MACSQuant,
                                            matched,
                                            stats = "mean"
      )
      # print(statistics[i, ])
      message(paste(i, "...", sep = ""))
    } else {
      warning("Outside of selection field, select your data again")
      break
    }
  }
  message("OK\n")
  sorted_matrix_final <- sorted_matrix
  if (i != number_of_conditions) {
    stop("Process interrupted please start again")
  }
  message("\n--> Done: replicates identified")
  message("--> Done: statistics on each condition replicates")


  if (control == TRUE) {
    MACSQuant@param.experiment$control <- TRUE
    message("...You can now select your control replicates...")
    message("--> 1 control: ...")

    # check with users
    sorted_matrix_ctrl <- matrix(
      data = NA, ncol = number_of_replicates,
      nrow = 1
    )

    loc2 <- locator(type = "n", n = number_of_replicates)
    if (length(loc2) == 2) {
      sorted_matrix_ctrl[1, ] <- to_well_names(loc2, 1, well_letter)
      matched <- match_id_line(MACSQuant, sorted_matrix_ctrl[1, ])
      if (length(matched) != number_of_replicates * 2) {
        warning(c(
          paste(
            "Your file may not contains 2 gates",
            "for one or more replicates in: ",
            sep = " "
          ),
          paste(sorted_matrix_ctrl[1, ], collapse = ","),
          "\nPlease",
          "be sure to only have P1 and P1/P2 gates in your file"
        ))
      }
      statistics[i + 1, ] <- compute_statistics(MACSQuant, matched,
                                                stats = "mean"
      )
      message("OK...")

      # print(paste('selection',paste(sorted_matrix_ctrl[1,],collapse =
      # ','),sep=' '))
      message("\n--> Done: statistics on each control replicates")
      sorted_matrix_final <- rbind(sorted_matrix_final,
                                   sorted_matrix_ctrl)
      message(sorted_matrix_final)
    } else {
      stop("Outside of selection field, select your data again")

    }
  }
  MACSQuant@my_replicates_sorted <- sorted_matrix_final
  if (control == FALSE &
      (length(MACSQuant@param.experiment$c_names) == 0)) {
    legend("topright",
           inset = c(-0.3, 0.05), legend = c("Conditions"),
           pch = c(1), col = col[1])
    MACSQuant@param.output$plt.labels <- c(rep("Conditions",
                                               number_of_conditions))
    plt.labels <- c(rep("Conditions", number_of_conditions))
    statistics[, seq(3, 7)] <- vapply(statistics[, seq(3, 7)], function(x) {
      as.numeric(x)
    }, numeric(number_of_conditions))
  } else if (control == TRUE &
             (length(MACSQuant@param.experiment$c_names) == 0)) {
    legend("topright", inset = c(-0.3, 0.05), legend = c(
      "Conditions",
      "Control"
    ), pch = c(1), col = c(col[1], 1))
    MACSQuant@param.output$plt.labels <- c(
      rep("Conditions",
          number_of_conditions), "Control")
    plt.labels <- c(rep("Conditions", number_of_conditions), "Control")
    statistics[, seq(3, 7)] <- vapply(statistics[, seq(3, 7)], function(x) {
      as.numeric(x)
    }, numeric(number_of_conditions + 1))
  } else if (control == FALSE &
             (length(MACSQuant@param.experiment$c_names) != 0)) {
    legend("topright",
           inset = c(-0.3, 0.05), legend = MACSQuant@param.experiment$c_names,
           pch = c(1), col = col
    )
    MACSQuant@param.output$plt.labels <-
      rep(MACSQuant@param.experiment$c_names)
    plt.labels <- rep(MACSQuant@param.experiment$c_names)
    statistics[, seq(3, 7)] <- vapply(statistics[, seq(3, 7)], function(x) {
      as.numeric(x)
    }, numeric(number_of_conditions))
  } else if (control == TRUE &
             (length(MACSQuant@param.experiment$c_names) != 0)) {
    legend("topright", inset = c(-0.3, 0.05), legend = c(
      MACSQuant@param.experiment$c_names,
      "Control"
    ), pch = c(1), col = c(col, 1))
    MACSQuant@param.output$plt.labels <-
      c(MACSQuant@param.experiment$c_names, "Control")
    plt.labels <- c(MACSQuant@param.experiment$c_names, "Control")
    statistics[, seq(3, 7)] <- vapply(statistics[, seq(3, 7)], function(x) {
      as.numeric(x)
    }, numeric(number_of_conditions + 1))
  }


  if (save.files == TRUE) {
    MACSQuant@param.output$save.files <- TRUE
    MACSQuant <- create_output_folder(MACSQuant)
    dev.copy(png, paste(MACSQuant@param.output$path,
                        "/outputMQ/plate_template.png", sep = ""),
             width = 600, height = 600,
             units = "px"
    )
    dev.off()
    message("--> Done: image saved")
    write.table(statistics, paste(MACSQuant@param.output$path,
                                  "/outputMQ/statistics.txt ", sep = ""),
                sep = "\t")
    message("--> Done: statistics table saved")
  }


  message("--> Done: replicates stored in variable my_replicates_sorted")
  MACSQuant@statistics <- statistics

  indices <- order_data(MACSQuant)
  MACSQuant@my_data_sorted <- MACSQuant@my_data[indices, ]

  if (MACSQuant@param.experiment$control == TRUE &
      length(MACSQuant@param.experiment$doses) != 0) {
    col <- c(col, 1)
  }  else if (MACSQuant@param.experiment$control == TRUE &
              length(MACSQuant@param.experiment$doses) == 0) {
    MACSQuant@param.experiment$doses <- seq_len(number_of_conditions + 1)
    col <- c(col, 1)
  } else if (MACSQuant@param.experiment$control == FALSE &
             length(MACSQuant@param.experiment$doses) == 0) {
    MACSQuant@param.experiment$doses <- seq_len(number_of_conditions)
  }

  MACSQuant@param.output$plt.col <- col
  if (length(MACSQuant@param.experiment$doses.alt) == 0) {
    for (flav in c("counts", "percent"))
    {
      p <- barplot_data(MACSQuant, col, plt.flavour = flav,
                        plt.labels = plt.labels, plt.combo = FALSE)
      grid.arrange(p)
      if (save.files == TRUE) {
        ggsave(paste(MACSQuant@param.output$path,
                     "/outputMQ/barplot_", flav, "_pipeline.png", sep = ""),
               width = 15.875, height = 15.875,
               units = "cm", p)
        message("--> Done: image saved")
      }
    }

  } else {
    message("...You can now run barplot_data(...plt.combo = TRUE)...")
  }
  return(MACSQuant)
}