label_cluster.R

#!/usr/bin/env Rscript
if (!require(optparse)) install.packages("optparse"); library(optparse)

option_list <- list(
  make_option(opt_str = c("-i", "--input"), default = NULL, help = "Input TSV-file. Output from tomtom", metavar = "character"),
  make_option(opt_str = c("-o", "--output"), default = NULL, help = "Output TSV-file.", metavar = "character")
)

opt_parser <- OptionParser(option_list = option_list, 
                           description = "Adding Cluster ID to Query_ID Column.",
                           epilogue = "Author: Rene Wiegandt <Rene.Wiegandt@mpi-bn.mpg.de>")

opt <- parse_args(opt_parser)

#' Adding Cluster ID to Query_ID Column
#' 
#' @param input <string> TSV-file. Output from tomtom.
#' @param input <string> Output name.
#'
#' @author René Wiegandt
#' @contact rene.wiegandt(at)mpi-bn.mpg.de
label_cluster <- function(input, output){
  #Reading TSV-file
  tsv <- data.table::fread(input, header = T, sep = "\t")
  
  #Getting cluster ID/number 
  splitted_name <- unlist(strsplit(input, "\\_|\\."))
  cluster_number <- as.numeric(splitted_name[length(splitted_name) - 1]) + 1
  
  #Adding cluster ID to first column
  tsv$Query_ID <- paste0(tsv$Query_ID, ".", cluster_number)
  
  
  data.table::fwrite(tsv, file = output, sep = "\t", col.names = FALSE)
}

# run function label_cluster with given parameteres if not in interactive context (e.g. run from shell)
if (!interactive()) {
  label_cluster(opt$input, opt$output)
}