generate_star_genomes.qsub

#!/bin/bash -l
#$ -S /bin/bash
#$ -cwd
#$ -j y
#$ -m eas
#$ -pe omp 8
#$ -N generate_star_genomes

# Build a set of STAR genomes for a given reference
# Adam Gower, based on scripts by Josh Campbell

# Parse command-line arguments
eval set -- "$(
  getopt --options=r:f:o:s: \
         --longoptions=reference:,fasta-path:,output-path:,star-version: \
         --name "$0" -- "$@"
)"

while true
do
  case "$1" in
    -r|--reference)
      reference="$2"
      shift 2 ;;
    -f|--fasta-path)
      fasta_path="$(readlink --canonicalize "$2")"
      shift 2 ;;
    -o|--output-path)
      output_path="$(readlink --canonicalize "$2")"
      shift 2 ;;
    -s|--star-version)
      star_version="$2"
      shift 2 ;;
    --)
      shift
      break ;;
    *)
      echo "Internal error"
      exit 1 ;;
  esac
done

if [[ ${reference} == "" ]]
then
  echo    "Usage:"
  echo    "  qsub [qsub flags] generate_star_genomes.qsub [options]"
  echo    "       -r|--reference [reference name]"
  echo    "       -f|--fasta-path [Path to FASTA files]"
  echo    "Options:"
  echo    "  -r, --reference       Reference name"
  echo    "                        (e.g., hg38, mm10, bosTau9, MesAur1.0, etc.)"
  echo    "  -f, --fasta-path      Path to FASTA input files"
  echo    "                        (i.e., path containing 'base' folder, etc.)"
  echo -n "  -o, --output-path     "
  echo -n "Path where output will be written "
  echo    "(will be created if it does not exist)"
  echo -n "                        "
  echo    "(Default: ./[STAR version]/[reference name]/)"
  echo    "  -s, --star-version    Version of STAR to use"
  echo    "                        (if omitted, default STAR version is used)"
else
  if [[ ${star_version} == "" ]]
  then
    # If no STAR version was specified, load the default module
    module load star
  else
    # Check whether a valid module version was specified
    # If so, load it; if not, load the default module
    if [[ $(module avail star/${star_version} |& wc -l) != 0 ]]
    then
      module load star/${star_version}
    else
      module load star
    fi
  fi
  # Get version of STAR module that was loaded
  star_version="$(STAR --version | sed 's/STAR_//')"

  # If output path is not specified, a directory named with the STAR version and
  # UCSC genome build will be created within the current working directory,
  # and output will be written there
  if [[ "${output_path}" == "" ]]
  then
    output_path="$(pwd)/${star_version}/${reference}"
  fi
  # Create the output path if it does not already exist
  [ ! -d "${output_path}" ] && mkdir --verbose -p "${output_path}"/ 
  # Ensure directory is not world-readable yet (to keep others out until ready)
  chmod 2700 "${output_path}"/

  echo "Reference name: ${reference}"
  echo "FASTA input files located in: ${fasta_path}/"
  echo "STAR version: ${star_version}"
  echo "STAR genomes will be written to: ${output_path}/"

  cd "${fasta_path}"/ || exit

  # Run STAR in 'genomeGenerate' mode for each FASTA file
  for type in *
  do
    # Create genome-level directory that is not world-readable yet
    # (to keep users out until it's ready)
    mkdir --verbose --mode=2700 "${output_path}"/${type}/
    # Generate genome
    # Note: '--outTmpDir $TMPDIR' throws an error;
    #       STAR needs to make the temporary directory itself
    STAR \
      --runMode genomeGenerate \
      --genomeDir "${output_path}"/${type} \
      --genomeFastaFiles "${fasta_path}"/${type}/${reference}.fa \
      --outFileNamePrefix "${output_path}"/${type}/ \
      --outTmpDir $TMPDIR/${type} \
      --runThreadN $NSLOTS
  done
  # Make all output files and directories world-readable and read-only
  chmod -Rc ugo-w,ugo+rX ${output_path}/
fi