Run CuReSim n times for each sequence

generation.sh

@@ -19,5 +19,5 @@ fastq=".fastq"
 filename="sequence"
 prefix="curesim_"
 
-Rscript src/repertoire.r "$sequences" "$sequencing_runs" && java -jar tools/CuReSim.jar -n "$sequencing_runs" -m "$read_mean_size" -sd "$read_variance_size" -f "$data_directory$filename$fasta" -o "$data_directory$prefix$filename$fastq"
+Rscript src/repertoire.r "$sequences" && java -jar tools/CuReSim.jar -n $((sequencing_runs * sequences)) -m "$read_mean_size" -sd "$read_variance_size" -f "$data_directory$filename$fasta" -o "$data_directory$prefix$filename$fastq"
 rm "$data_directory/log.txt"

src/repertoire.r

@@ -10,9 +10,7 @@ generate_repertoire <- function(number_of_sequences) {
   ))
 }
 
-process_data <- function(repertoire, reads) {
+process_data <- function(data) {
-  columns <- c("sequence", "v_call", "j_call")
-  data <- repertoire[, columns]
   dna_sequence <- Biostrings::DNAStringSet(data$sequence)
   data$sequence <- Biostrings::reverseComplement(dna_sequence)
   names(data$sequence) <- paste(rownames(data), data$v_call, data$j_call)
@@ -21,12 +19,12 @@ process_data <- function(repertoire, reads) {
 
 parse_cli_arguments <- function() {
   args <- commandArgs(trailingOnly = TRUE)
-  if (length(args) != 2) {
+  if (length(args) != 1) {
-    stop("usage: repertoire.r <number of sequences> <sequencing runs>")
+    stop("usage: repertoire.r <number of sequences>")
   }
-  return(c(args[1], args[2]))
+  return(args[1])
 }
 
-args <- parse_cli_arguments()
+argument <- parse_cli_arguments()
-repertoire <- generate_repertoire(number_of_sequences = as.integer(args[1]))
+repertoire <- generate_repertoire(number_of_sequences = as.integer(argument))
-process_data(repertoire = repertoire, reads = as.integer(args[2]))
+process_data(data = repertoire)