R/timsDataLoader.R

In TMPierpont/timsBasicHeatmap: Tools for analyzing manually gated flow data

#' This grabs flowjo files and extracts data
#' I could make this combine the files, but it'll be easier to use and more versatile if we just combine them outside this function
#' @export
flowJoDataLoader <- function(
sample_list, #sample info reference list
file, #flowjo File
fcs_files, #directory with FCS files
sample_info = NULL, #The columns that are not counts or samples that are to be transferred from the sample info sheet
main_event = NULL, #The population you want to treat as the main population when calculating totals.
debug=FALSE,
group=NULL,
counts="counts",#This is the column with cell counts in your sample list per ml
volumes="volumes", #This is the column with total volumes of your cell count
samples="sample"#this is the column with your sample IDs in your sample list
) {
  if (debug) {cat("\nopening wsp file")}
  ws <- open_flowjo_xml(file)

  if (debug) {cat("... good!\nLoading fcs files")}
  gs <- flowjo_to_gatingset(ws, name = 1, path = fcs_files)
  nodelist <- gs_get_pop_paths(gs, path = "full")
  shortnodeList  <- gs_get_pop_paths(gs, path = "auto")

  #remove any column manes that are NA or are actually called NA (excel imports do that sometimes)
  if (any(is.na(colnames(sample_list)))) { #Check before doing this one or it'll cause an error
    sample_list <- sample_list[-which(is.na(colnames(sample_list)))]  
  }
  if (any(is.na(colnames(sample_list)))) { #Check before doing this one or it'll cause an error  
    sample_list <- sample_list[-which(colnames(sample_list)=="NA")]   
  }
  ######################
  #match the sample file to the myData DF
  ######################
  myData <- as.data.frame(gs_pop_get_stats(gs, nodes = nodelist[1], xml = TRUE, type="percent")[,1])
  colnames(myData)[1] <- "sample"

  if (debug) {print("cleaning up wsp file IDs")}  
  #first clean up the fcs file names
  for (i in 1:nrow(myData)) {
    myData[i,'sample'] <- str_split(myData[i,'sample'],"\\.",simplify = TRUE)[1]
  }
  
  if (debug) {cat("... good!\nremoving any '(1)' from sample list")}  
  #myData[samples] <- str_remove(myData[samples], "\\(1\\)") #sometimes if a file is saved twice, the fcs file gets a (1) and it's a pain in the...
  
  #Check to see if the sample IDs are numeric, and if they are, then convert them to string and add 00s as appropriate to equalize the characters used
  if (debug) {cat("... good!\nChecking if sample IDs are numeric, converting if necassary\n")}   
  if (apply(sample_list[samples], 2, is.numeric)) {
    max_sample <- nchar(toString(max(sample_list[samples]))) #get the biggest number, convert it to string, and then count the characters
    temp_list <- unlist(as.list(sample_list[samples])) #yes... I know I made it a list and unlisted, but somehow that's the only way I could find to get the desired type
    sample_list[samples] <- formatC(temp_list, width=max_sample, flag="0") #then convert them to 00x format as characters
    temp_list <- as.numeric(unlist(as.list(myData[samples])))
    myData[samples] <- formatC(temp_list, width=max_sample, flag="0") #same for FCS sample IDs
  } else {
    print(myData$sample)
    #if (grepl("(1)", myData$sample)) {myData$sample <- str_remove(myData$sample, "\\(1\\)")} #sometimes if a file is saved twice, the fcs file gets a (1) and it's a pain in the...
  }
  
  #Making sure sample names are unique
  if (any(duplicated(sample_list[samples]))) {warning("You have duplicate sample IDs in your sample list!")}
  if (any(duplicated(myData[samples]))) {warning("You have duplicate sample IDs in your WSP files!")}
  if (debug) {cat("... good!\nPulling sample list info from into new dataframe and combining it with cleaned IDs from the WSP file\n")}   
  
  #make the columns if they need to be made
  if(any(str_detect(colnames(sample_list), counts))){myData$counts <- ""}
  if(any(str_detect(colnames(sample_list), volumes))){myData$volumes <- ""}
  
    for (i in 1:nrow(myData)) {
      if (debug) {cat(paste("\r    Hunting for:",myData[i,'sample']))}
    replaceNum <- stringr::str_which(unlist(as.list(sample_list[samples])), paste0(myData[i,'sample']))
    if(any(str_detect(colnames(sample_list), counts))){
       myData[i,'counts'] <- sample_list[replaceNum,counts]
      }
    if(any(str_detect(colnames(sample_list), volumes))){
      myData[i,'volumes'] <- sample_list[replaceNum,volumes]
      }
    for (a in 1:length(sample_info)) {
        myData[i, sample_info[a]] <- sample_list[replaceNum, sample_info[a]]
      }
    }

  #########################
  #Now load all the numbers
  #########################
  
  if (debug) {cat("... good!\nChecking to if we should use another pop instead of root")}   
  #If someone wants to use something other than root as their main population to normalize to:
  if (is.null(main_event)) {root <- nodelist[1]} else { 
    if (debug) {cat(paste("\nUsing", main_event, "instead"))}   
    root <- main_event
    if (!any(nodelist == main_event)) { 
      print(paste("failed to find population", main_event, "only found:", nodelist))
      return(NULL)
    }
  }
  
  if (debug) {cat("... good!\nGenerating Totals and applying totals")}   
  #Totals first
  myData <- as.data.frame(myData)
  
  if(any(str_detect(colnames(sample_list), volumes))){ 
    myData['counts'] <- as.numeric(myData$volumes)*as.numeric(myData$counts)
  } else {warning("No volumes found, using counts only")}
  
  for (i in 2:length(nodelist)) {
    myData['new_col'] <- gs_pop_get_stats(gs, nodes = nodelist[i])[,3]
    if(any(str_detect(colnames(sample_list), counts))){
      #Get % total by dividing by total events ("Root")
      myData['new_col'] <- myData['new_col'] / as.data.frame(gs_pop_get_stats(gs, nodes = root, type="count")[,3])
      
      #Multiply by total counts
      
      myData$new_col <- myData$new_col * as.numeric(myData$counts)
      
      colnames(myData)[ncol(myData)] = toString(paste("Total", shortnodeList[i]))
    }else{
      colnames(myData)[ncol(myData)] = toString(paste("Events", shortnodeList[i]))      
    }
  }
  if (debug) {cat("... good!\nApplying percents")}   
  #Then Percentages
  for (i in 2:length(nodelist)) {
    myData['new_col'] <- gs_pop_get_stats(gs, nodes = nodelist[i], type="percent")[,3]
    colnames(myData)[ncol(myData)] = toString(paste("Percent", shortnodeList[i]))
  }
  colnames(myData)[colnames(myData) == "counts"] <- counts
  colnames(myData)[colnames(myData) == "volumes"] <- volumes
  
  if (debug) {cat("... good!\nDone! returning data!")}
  return(myData)
}