R/dimsum__check_countfile.R

In lehner-lab/DiMSum: A pipeline for processing deep mutational scanning data

#' dimsum__check_countfile
#'
#' Check whether user-specified count file correctly formatted.
#'
#' @param dimsum_meta an experiment metadata object (required)
#' @param return_data whether or not to return data (default: FALSE)
#'
#' @return Reformatted data.table
#' @export
#' @import data.table
dimsum__check_countfile <- function(
  dimsum_meta,
  return_data = FALSE
  ){

  ### Abort if no count file supplied
  if(is.null(dimsum_meta[["countPath"]])){return(NULL)}

  ### Load variant data
  input_dt <- data.table::fread(dimsum_meta[["countPath"]])

  ### Nucleotide sequence checks (nt_seq column)
  #Check if mandatory columns present
  mandatory_cols <- c("nt_seq")
  if(sum(unlist(lapply(mandatory_cols, "%in%", colnames(input_dt)))==FALSE)!=0){
    stop(paste0("One or more mandatory columns missing from file specified by countPath ('nt_seq')"), call. = FALSE)
  }
  #Check nucleotide sequence column is of type character 
  if(sapply(input_dt, typeof)["nt_seq"]!="character"){
    stop("One or more invalid 'nt_seq' values in variant count file specified by 'countPath'. Only valid nucleotide sequences allowed (A/C/T/G).", call. = FALSE)
  }
  #Set nucleotide sequence to lower case
  input_dt[, nt_seq := tolower(nt_seq)]
  #Check nucleotide sequences are valid (ACGT characters only)
  if(sum(input_dt[,grepl("[^acgt]", nt_seq)])!=0){
    stop("One or more invalid 'nt_seq' values in variant count file specified by 'countPath'. Only valid nucleotide sequences allowed (A/C/T/G).", call. = FALSE)
  }

  ### Count column checks 
  #Check if sample name columns present
  mandatory_cols <- unique(dimsum_meta[["exp_design"]][,"sample_name"])
  if(sum(unlist(lapply(mandatory_cols, "%in%", colnames(input_dt)))==FALSE)!=0){
    stop(paste0("One or more sample names in experimentDesign file missing from column names in variant count file specified by 'countPath'"), call. = FALSE)
  }
  #Check all count columns are of type integer 
  typeof_cols <- sapply(input_dt[,.SD,,.SDcols = names(input_dt)[names(input_dt)!="nt_seq"]], typeof)
  if(sum(typeof_cols!="integer")!=0){
    stop(paste0("Invalid type of sample count column in variant count file specified by 'countPath'. Only positive integers allowed (zero inclusive)."), call. = FALSE)
  }
  #Check all count columns positive integer zero inclusive
  if(input_dt[,min(.SD, na.rm = T),,.SDcols = names(input_dt)[names(input_dt)!="nt_seq"]]<0){
    stop(paste0("Invalid type of sample count column in variant count file specified by 'countPath'. Only positive integers allowed (zero inclusive)."), call. = FALSE)
  }

  ### Duplicated variants check
  if(input_dt[,sum(duplicated(nt_seq))]!=0){
    stop(paste0("Duplicated 'nt_seq' values not allowed in variant count file specified by 'countPath'."), call. = FALSE)
  }

  ### Return data if required
  if(return_data){
    #Sample names (ignore 'technical_replicate' column)
    sample_names <- as.list(paste0(
      dimsum_meta[["exp_design"]][,"sample_name"], '_e', 
      dimsum_meta[["exp_design"]][,"experiment"], '_s', 
      dimsum_meta[["exp_design"]][,"selection_id"], '_b', 
      dimsum_meta[["exp_design"]][,"biological_replicate"], '_tNA_count', sep = ""))
    names(sample_names) <- dimsum_meta[["exp_design"]][,"sample_name"]

    #Reformat count column names
    names(input_dt)[names(input_dt)!="nt_seq"] <- unlist(sample_names[names(input_dt)[names(input_dt)!="nt_seq"]])

    return(input_dt)
  }
}