R/pre_filter.R

In kgellatl/SignallingSingleCell: Network Analysis for single cell RNA sequencing Data (sc-RNASeq)

#' Filter Data
#'
#' This function will filter the data to remove cells and/or genes
#'
#' @param input the input ex_sc
#' @param minUMI min number of UMIs per cell
#' @param maxUMI max number of UMIs per cell
#' @param threshold UMI threshold for gene detection
#' @param minCells number of cells expressed above threshold for a given gene
#' @param print_progress will print progress if TRUE
#' @export
#' @details
#' When processing the data, low quality cells may contain very few UMIs,
#' while some overrepresented cell barcodes may indicate barcode bleedover or cell doublets.
#' Filtering out both low and high UMI count cells is recommended before normalization.
#' @examples
#' filtered_data <- filter_UMIs(input = ex_sc_example, minUMI = 1000, maxUMI = 10000, threshold = 1, minCells = 10, print_progress = TRUE)

pre_filter <- function(input, minUMI = 0, maxUMI = NA, threshold=NA, minCells=NA, print_progress = TRUE){
  if(is.na(threshold) != TRUE){
    if(is.na(minCells) != TRUE){
    }
    if(print_progress == TRUE){
      print("Filtering Genes")
    }
    gCount <- apply(exprs(input),1,function(x) length(which(x>=threshold)))
    input <- input[(which(gCount >= minCells)),]
  }
  if(print_progress == TRUE){
    print("Filtering Low Cells")
  }
  cSum <- apply(exprs(input),2,sum)
  if(is.na(maxUMI)){
    maxUMI <- max(cSum)
  }
  low <- which(cSum < minUMI)
  high <- which(cSum > maxUMI)
  remove <- c(low, high)
  if(length(remove) > 0){
    input <- input[,-remove]
  }
  return(input)
}