#'@title Export Strongest Correlated Regions
#'
#' @description This function identifies a user-specified number of regions where expression correlates most
#' strongly and positively to a user-specified brain region.
#' If there are any brain regions where expression correlates negatively with expression in that region, this
#' function also finds and ranks those in a separate column.
#' This function then exports a .csv file that can be fed into the Matlab function CustomBrainRegionStack
#' so that the anatomical organization of these brain regions can be visualized in 3D. Regions with positive
#' scores will be colored green while regions with negative scores will be purple.
#' This package requres you to have the janitor, readxl, and tidyverse packages loaded.
#' @param corr_object This is the unmodified output of your cor() function.
#' @param brain_region This is the brain region you want to examine. Be sure to enter the region name exactly
#' as it appears in your correlation object. For example, type out rhombencephalon_qrfp_neuron_cluster_sparse.
#' @param num2show This is the max number of regions in EACH color you want displayed on the graph. You may
#' get fewer (for example if there are no regions negatively correlated with your region of choice)
#' @param filtered this is whether you filtered out brain regions due to alignment problems. Enter "none" if
#' you filtered out no regions (your corr matrix is 293 x 293). Enter "spinalcord" if you filtered
#' out the spinal cord (your corr matrix is 275 x 275). Enter "rhomb7" if you filtered out everything caudal
#' to rhombomere 7 (your corr matrix is 254 x 254).
#' @export
#' @examples


exportStrongestCorrelatedRegions <- function(corr_object, brain_region, num2show, filtered) {

  num2show = as.numeric(num2show)

  # importing correct list of brain regions
  if (filtered == "rhomb7") {
    regions = read_xlsx("rhomb7_filtered_regions_underscores.xlsx", col_names = c("ROIname"))
  } else if (filtered == "spinalcord") {
    regions = read_xlsx("spinalcord_filtered_regions_underscores.xlsx", col_names = c("ROIname"))
  } else if (filtered == "none") {
    regions = read_xlsx("unfiltered_regions_underscores.xlsx", col_names = c("ROIname"))
  } else {
    print("For the 'filtered' parameter, please enter 'none' or 'spinalcord' or 'rhomb7'.")
  }

  corr_wholebrain_tib <- as_tibble(corr_object)

  myROI_corr <- corr_wholebrain_tib %>%
    select(brain_region)

  myROI_corr_regions <- cbind(regions, myROI_corr)
  colnames(myROI_corr_regions) <- c("ROIname","corr")

  myROI_corr_ranked <- myROI_corr_regions %>%
    arrange(desc(corr))

  colnames(myROI_corr_ranked) <- c("ROIname","corr")

  top_ranked_regions <- slice_head(myROI_corr_ranked, n = (num2show+1)) %>% #num2show +1 because top region will be the ROI itself
    filter(corr< .999999999) #filter out the brain region itself, which will have a correlation of 1.0

  bottom_ranked_regions <- slice_tail(myROI_corr_ranked, n = num2show)
  bottom_regions_filtered <- bottom_ranked_regions %>% # only include bottom regions if their scores are actually <0 (anticorrelated)
    filter(corr < 0) %>%
    mutate(corr = abs(corr)) #...but Matlab needs positive numbers to graph

  brain_reg_columnar <- read_csv("293_brain_regions_formatted.csv") # THIS ONLY IS COMPATIBLE WITH THE MATLAB PROGRAM IF YOU USE THE FULL 293 BRAIN REGIONS
  colnames(brain_reg_columnar) <- "ROIname"

  posJoin <- left_join(brain_reg_columnar, top_ranked_regions)  # fills in correlations at corresponding brain regions
  posJoin[is.na(posJoin)] = 0 # replaces NAs with 0s
  colnames(posJoin) <- c("ROIname","PosCorr")

  negJoin <- left_join(brain_reg_columnar, bottom_regions_filtered, by = "ROIname") # fills in scores at corresponding brain regions
  negJoin[is.na(negJoin)] = 0 # replaces NAs with 0s
  colnames(negJoin) <- c("ROIname","NegCorr")

  posJoin$NegCorr <- negJoin$NegCorr  #adds negative value column to main object

  correlations <- posJoin %>%
    select(PosCorr,NegCorr)  # gets rid of ROI names because they make Matlab unhappy

  reg_name <- as.character(brain_region)

  filename <- paste(reg_name, "Corr.csv", sep = "") #descriptive filename

  write_csv(correlations, filename) #write csv

  all_ranked_regions <- rbind(top_ranked_regions, bottom_ranked_regions)
  return(all_ranked_regions)

}