#' @title Normalize signal by...
#'
#' @description This function helps compensate for differences in staining ("collected"), imaging parameters ("imaged"), or parents ("pair")
#' @param data a tibble containing the standard metadata + individual fish as the columns and brain region names as the rows.
#' #' If you use the output from tidyImportedData or tidyImportedDataUnderscore you should be all set.
#' @param byColumn This is the column that the function will group and normalize by. Essentially, every fish that
#' #' has the same value for that column will be averaged together, then all the signal values for each fish
#' #' will be divided by the average overall signal for the group to which it belongs.
#' @export
#' @examples

normalizeSignalBy <- function(data, by_column) {

  #by_column should be "collected" or "imaged" or "pair"

  #note: I tested this function against my original script output-- outputs were identical
  # I tested my original script output against an Excel spreadsheet where I did all the manipulations as I intended

  studyDesign <- data[, 1:7]

  dataOnly <- data[,8:length(data)]

  brainRegions <- colnames(dataOnly)

  numericDataOnly <- dataOnly %>%
    mutate_all(as.numeric)

  numericDataOnly$avgSig <- rowMeans(numericDataOnly)

  data2 <- cbind(studyDesign, numericDataOnly)

  if (by_column == "collected") {
    data2 <- data2 %>%
      group_by(collected)
      meanSigByGrouped <- summarise(data2, mean(avgSig))
      obj <- left_join(data2, meanSigByGrouped, by = "collected")
  }

  else if (by_column == "imaged") {
    data2 <- data2 %>%
      group_by(imaged)
    meanSigByGrouped <- summarise(data2, mean(avgSig))
    obj <- left_join(data2, meanSigByGrouped, by = "imaged")
  }

  else if (by_column == "pair") {
    data2 <- data2 %>%
      group_by(pair)
    meanSigByGrouped <- summarise(data2, mean(avgSig))
    obj <- left_join(data2, meanSigByGrouped, by = "pair")
  }

  else {
    print("Please enter by_column = collected or imaged or pair (in quotes)")
  }

  obj2 <- obj[,8:length(obj)]

  obj3 <- obj2 %>% sapply(`/`, obj[,length(obj)])
  obj3[is.na(obj3)] = 0

  obj3 <- as_tibble(obj3)

  obj4 <- obj3[1:(length(obj3)-2)]
  colnames(obj4) <- brainRegions
  obj4 <- cbind(studyDesign,obj4)

  obj4 <- as_tibble(obj4)

  return(obj4)

  write_csv(obj4, "normalizedSignal_Test1_210518.csv")

}


#### testing the function
#setwd("C:/Users/Joe/Documents/MATLAB/WholeBrainExpression")

#original_raw_data <- read_csv("data/AllFullBrains_Together_210412.csv")

#tidiedData <- tidyImportedDataUnderscore(original_raw_data)

#studyDesign <- tidiedData[, 1:7]

# dataOnly <- tidiedData[,8:length(tidiedData)]
#
# numericDataOnly <- dataOnly %>%
#   mutate_all(as.numeric)
#
# numericData <- cbind(studyDesign, numericDataOnly)
#
# result <- normalizeSignalBy(numericData, by_column = "collected")
#
# write_csv(result, "normalizedSignal_Test3Function_210518.csv")