locigenesis/src/repertoire.r

library(immuneSIM)
library(Biostrings)

generate_repertoires <- function(number_of_sequences) {
  b_chain <- immuneSIM(
    number_of_seqs = number_of_sequences,
    species = "hs",
    receptor = "tr",
    chain = "b",
    verbose = TRUE
  )
  return(b_chain)
}

# TODO save also v_call and j_call
preprocess_data <- function(repertoire, sequencing_runs) {
  sequences <- as.character(repertoire$sequence)
  reads <- Biostrings::DNAStringSet(rep(sequences, sequencing_runs))
  names(reads) <- seq_len(length(reads))
  reverse_complement <- Biostrings::reverseComplement(reads)
  return(reverse_complement)
}

save_data <- function(repertoire) {
  file_name <- "data/sequence.fastq"
  # TODO Change format to fasta
  Biostrings::writeXStringSet(repertoire, file_name, format = "fastq")
}

parse_cli_arguments <- function(args) {
  if (length(args) != 2) {
    stop("usage: repertoire.r <number of sequences> <sequencing_runs>")
  }
  return(c(args[1], args[2]))
}

args <- commandArgs(trailingOnly = TRUE)
parameters <- parse_cli_arguments(args)
number_of_sequences <- as.integer(parameters[1])
sequencing_runs <- as.integer(parameters[2])
repertoire <- generate_repertoires(number_of_sequences)
processed_data <- preprocess_data(repertoire, sequencing_runs)
save_data(processed_data)