run_assembly.uge-nextflow

#!/usr/bin/env bash
#$ -cwd

# This wrapper script manages job submission of
#  the wf-paired-end-illumina-assembly workflow
#  to the UGE scheduler.
SCRIPT_NAME="$(basename ${0#_} .uge-nextflow)"

# Define script usage
usage() {
  echo "
Usage: ${0##*/} InputDirectory [OutputDirectory] [-h|--help]

Required:
  <InputDirectory>   Path containing PE Illumina 1.8+ files as *.fastq.gz

  Sample names are extracted from the first underscore of read files. Hyphens,
  periods, and commas are automatically discarded from the name to avoid errors.
  WARNING: If you have HiSeq data where samples are ran on more than one lane,
  be sure to concatenate them.

  File searching goes 2 levels deep to find .fq and .fastq file pairs,
  optionally gunzip compressed.

  Sample pairing of sister reads is done by consecutively using filenames sorted
  with \`ls\`, which flexibly allows for naming other than traditional R1 and R2.
  Verify that files will properly pair by checking the output order of \`ls\`
  within the <InputDirectory>.

Optional:
  -h | --help         Show this help message and exit.
  <OutputDirectory>   Location for the output files, which
                      includes job logfiles. Output files
                      will be overwritten if already present.
                      Default: current working directory.
  "
}

# Color output for success, warnings, and errors to catch user's attention
GREEN_TXT='\033[0;32m'
RED_TXT='\033[0;31m'
YELLOW_TXT='\033[0;93m'
RED_BG='\033[41m'
COLOR_OFF='\033[0m'

# Allow user to re-specify the output path
prompt_new_outdir() {
  echo -e "${RED_BG}"
  read -p "Would you like to use a different output path? (yes|no) " -n 1 -r
  echo -e "${COLOR_OFF}"

  if [[ $REPLY =~ ^[yY] ]]; then
    echo -e "${RED_BG}"
    read -p "Enter new Output path: " _new_outdir
    echo -e "${COLOR_OFF}"

    OUT="$(readlink -f ${_new_outdir})"
  else
    echo -e "${RED_TXT}\nYou can re-run this workflow with a new output path."
    echo -e "Submission was cancelled.\n ${COLOR_OFF}"
    exit 0
  fi
}

# Check if there is a possibility data will be overwritten
prompt_if_previous_nextflow_run() {
  # Parameters:
  #  $1 = file path
  #  $2 = filename (no path) to search for
  if [ -f $(find "${1}" -maxdepth 1 -name "${2}*" | grep -q '.') ]; then
    echo -e "${YELLOW_TXT}\nThis workflow has been ran in output path before."
    echo -e "Processing samples will overwrite older data."
    echo -e "If you have jobs in queue this could be confusing.\n ${COLOR_OFF}"

    echo -e "${RED_BG}"
    read -p "Do you want to continue? (yes|no) " -n 1 -r
    echo -e "${COLOR_OFF}"

    if [[ $REPLY =~ ^[yY] ]]; then
      # Continue submission if user wants to possibly overwrite data
      echo -e "${YELLOW_TXT}\nData in ${1} will be overwritten."
      echo -e "Continuing to submit samples...${COLOR_OFF}"

      # Rename .work directory
      # if [[ -d "${OUT}/.work" ]]; then
      #   mv "${OUT}/.work" "${OUT}/.work_$(basename $(readlink -f ${OUT}/.work))"
      # fi
    else
      # If user doesn't want to continue, ask for a different output path
      prompt_new_outdir
      prompt_if_previous_nextflow_run "${OUT}/pipeline_info" "nextflow_log"
    fi
  fi
}

# Check argument requirements
[[ $1 == "--help" || $1 == "-h" ]] && { usage; exit 0; }
if [[ $# -lt 1  || $# -gt 2 ]]; then
  echo -e "${RED_TXT}ERROR:${COLOR_OFF} improper number ("$#") of arguments provided" >&2
  usage
  exit 1
fi

# Confirm we are on a host that can submit jobs
if [ ${HOSTNAME%%.*} == 'biolinux' ] || \
  [ ${HOSTNAME%%.*} == 'login02' ] || \
  [ ${HOSTNAME%%.*} == 'rosalind01' ] || \
  [ $(echo ${HOSTNAME%%.*} | sed 's/[0-9]//g') == 'node' ]; then
  :
else
  echo -e "${RED_TXT}ERROR:${COLOR_OFF} must be on aspen or biolinux" >&2
  exit 1
fi

# I/O handling
if [[ $1 =~ .+[[:space:]].+ ]]; then
  echo -e "${RED_TXT}ERROR:${COLOR_OFF} whitespace in $1 path unsupported" >&2
  exit 1
fi
if [[ $2 =~ .+[[:space:]].+ ]]; then
  echo -e "${RED_TXT}ERROR:${COLOR_OFF} whitespace in $2 path unsupported" >&2
  exit 1
fi
IN=$(readlink -f "$1")
if [[ -z "$2" ]]; then
  OUT="${PWD}"
else
  OUT=$(readlink -f "$2")
fi

# Get LAB_HOME or custom tmp/cache variables from user's ~/.bashrc,
#  while still enabling fancy people to override these from their
#  current working environment too.
source ${HOME}/.bashrc
if [[ -z "${LAB_HOME}" ]]; then
  echo -e "${RED_TXT}ERROR:${COLOR_OFF} ${LAB_HOME} not set" >&2
  exit 1
fi

# Check if Singularity environment variables are set,
#  and confirm the user has write access.
#  We expect these 2 paths to already exist.
if [[ -z ${SINGULARITY_CACHEDIR} ]]; then
  echo -e "${RED_TXT}ERROR:${COLOR_OFF} \$SINGULARITY_CACHEDIR not set" >&2
  exit 1
else
  if [[ ! -d ${SINGULARITY_CACHEDIR} ]] || [[ ! -w ${SINGULARITY_CACHEDIR} ]]; then
    echo -e "${RED_TXT}ERROR:${COLOR_OFF} \$SINGULARITY_CACHEDIR ${SINGULARITY_CACHEDIR} not writeable for ${USER}" >&2
    exit 1
  fi
fi
if [[ -z ${SINGULARITY_TMPDIR} ]]; then
  echo -e "${RED_TXT}ERROR:${COLOR_OFF} \$SINGULARITY_TMPDIR not set" >&2
  exit 1
else
  if [[ ! -d ${SINGULARITY_TMPDIR} ]] || [[ ! -w ${SINGULARITY_TMPDIR} ]]; then
    echo -e "${RED_TXT}ERROR:${COLOR_OFF} \$SINGULARITY_TMPDIR ${SINGULARITY_TMPDIR} not writeable for ${USER}" >&2
    exit 1
  fi
fi

# If Windows-style mount point (e.g., "Z:\Streptobacillus\Raw_FQs") given as a
#  path, check both user's home, lab home, and group as possible mount areas
#  to determine if files exist in 1 but not the others.
if [[ ${IN} =~ ^.+[A-Z]\:.+ ]]; then
  path_slashes_replaced="$(echo "${IN}" | cut -d ':' -f 2 | sed 's/\\/\//g')"
  user_home="${HOME}"/"${path_slashes_replaced}"
  user_group_main="/scicomp${path_slashes_replaced}" # Dropped slash between scicomp and var
  user_group_branch="/scicomp/groups/OID/NCEZID/DHCPP/BSPB/${path_slashes_replaced}"
  lab_home="${LAB_HOME}"/"${path_slashes_replaced}"
  paths=(
    "${user_home}"
    "${user_group_main}"
    "${user_group_branch}"
    "${lab_home}"
    )
  for path in "${paths[@]}"; do
    cnt_read_files=$(find -L "${path}" \
      -maxdepth 2 \
      -type f \
      -readable \
      -regextype posix-extended \
      -regex ".+_(R)?(1|2)(.+)?\.(fq|fastq)($|\.gz$)" \
      2> /dev/null \
      | wc -l)
    if [[ ${cnt_read_files} -ge 1 ]]; then
        echo -e "${YELLOW_TXT}\n${cnt_read_files} FastQ files found in: ${path}${COLOR_OFF}"
        echo -e "${RED_BG}"
        read -p "Use ${path} as Input Directory? (yes|no) " -n 1 -r
        echo -e "${COLOR_OFF}"
        if [[ $REPLY =~ ^[yY] ]]; then
          IN="${path}"
        else
          exit 0
        fi
        break
    fi
  done
fi

# For now, just output into shared LAB_HOME area
if [[ ${OUT} =~ ^.+[A-Z]\:.+ ]]; then
  echo -e "${YELLOW_TXT}\nWindows hard mount path detected as Output Directory ${COLOR_OFF}"
  proposed_outdir="${LAB_HOME}"/"$(date '+%Y-%b-%d_%a_%H:%M:%S')"

  echo -e "${RED_BG}"
  read -p "Use ${proposed_outdir} as Output Directory? (yes|no) " -n 1 -r
  echo -e "${COLOR_OFF}"

  if [[ $REPLY =~ ^[yY] ]]; then
    OUT="${path}"
  else
    echo -e "${RED_TXT}\nOkay, bad autoselection? Re-run with a different specified path ${COLOR_OFF}\n"
    exit 0
  fi
fi

# If Windows-style path from Core Facility given via email
#  (e.g., "\\...cdc.gov\groups\OID\...BCFB\by-instrument\NovaSeq\")
#  is given as path, form the Linux path
if [[ ${IN} =~ ^.+\.cdc\.gov\\.+\\.+ ]]; then
  inpath_stripped_hostname="${IN/*.cdc\.gov/\scicomp}"
  inpath_dewindowsed="$(echo "${inpath_stripped_hostname}" | sed 's/\\/\//g')"
  IN="${inpath_dewindowsed}"
fi

# Check for specific nextflow logfile for this workflow, which
#  means this nextflow workflow has been ran before in the specified outdir.
if [ -d "${OUT}/pipeline_info" ] && \
  [ -f $(find "${OUT}" -maxdepth 2 -name "nextflow_log*" | grep -q '.') ]; then
  prompt_if_previous_nextflow_run "${OUT}/pipeline_info" "nextflow_log"
fi

# Also check for unsupported behavior combining legacy bash and newer nextflow
if [ -d "${OUT}/.log" ] && \
  [ -f "${OUT}/.log/trim.asm.annot.job_ids.txt" ] && \
  [ -f "${OUT}/.log/trim.asm.annot.log.txt" ]; then
  echo -e "${RED_TXT}Legacy bash workflow detected in ${OUT}"
  echo -e "Combining data from that and this nextflow pipeline is unsupported."
  echo -e "Use a different output path. ${COLOR_OFF}"
  exit 1
fi

# Add found files to an array
found_files=( $(find -L ${IN} \
      -maxdepth 2 \
      -type f \
      -readable \
      -regextype posix-extended \
      -regex ".+_(R)?(1|2)(.+)?\.(fq|fastq)($|\.gz$)" | sort --dictionary-order ))

# Require even input file count for paired-end analysis
if [ ${#found_files[@]} -lt 1 ]; then
  echo -e "${RED_TXT}ERROR:${COLOR_OFF} no read sets found" >&2
  exit 1
fi
if [[ $((${#found_files[@]} % 2)) -ne 0 ]]; then
  echo -e "${RED_TXT}ERROR:${COLOR_OFF} uneven number (${#found_files[@]}) of FastQ files" >&2
  exit 1
fi

# Pair up the FastQ files based on basename (prefix) to submit jobs
submitted=()
for (( i=0; i<${#found_files[@]}; i+=2 )); do
  read1=${found_files[i]}
  read2=${found_files[i+1]}
  b1=$(basename ${read1} | cut -d _ -f 1 | sed 's/[-\.,]//g')
  b2=$(basename ${read2} | cut -d _ -f 1 | sed 's/[-\.,]//g')
  if [[ "${submitted[*]}" =~ " ${b1} " ]]; then
    echo -e "${RED_TXT}ERROR:${COLOR_OFF} ${b1} sample name occurs in >1 paired set" >&2
    exit 1
  fi
  if [ "${b1}" != "${b2}" ]; then
    echo -e "${RED_TXT}ERROR:${COLOR_OFF} improperly paired ${b1} ${b2}" >&2
    exit 1
  fi
  submitted+=(${b1})
done

# Prompt user if >10 samples to prevent accidentally assembling too many
if [[ ${#submitted[@]} -ge 10 ]]; then
  echo -e "${YELLOW_TXT}\nWARNING: ${#submitted[@]} samples have been identified:\n${COLOR_OFF}"
  for f in ${submitted[@]}; do
    echo -e "${YELLOW_TXT}${f}${COLOR_OFF}"
  done | xargs -L3 | column -t

  # Prompt user to verify submission of large sample set
  echo -e "${RED_BG}"
  read -p "Do you want to continue submitting ${#submitted[@]} samples? (yes|no) " -n 1 -r
  echo -e "${COLOR_OFF}"
  if [[ $REPLY =~ ^[yY] ]]; then
    echo -e "\nContinuing to submit ${#submitted[@]} samples...\n"
  else
    echo -e "${RED_TXT}\nSubmission cancelled ${COLOR_OFF}"
    exit 0
  fi
fi

# Set up log directory in OUT directory
mkdir -p ${OUT}/pipeline_info

# Get node number - <=230 = biolinux, >=231 = rosalind
NODE_NUM=$(echo ${HOSTNAME%%.*} | sed 's/node//1')

# If FastQ files are able to be submitted, submit and display basenames
if [[ ${#submitted[@]} -ge 1 ]] && \
  [[ ${HOSTNAME%%.*} == 'biolinux' || 'login02' ]] || \
  [[ ${NODE_NUM} -le 230 ]]; then
  # Submit Nextflow pipeline to Aspen HPC
  qsub \
    -m ba \
    -q all.q \
    -v IN=${IN} \
    -v OUT=${OUT} \
    -o ${OUT}/pipeline_info \
    -e ${OUT}/pipeline_info \
    -M ${USER}@cdc.gov \
    -v LAB_HOME=${LAB_HOME} \
    -N ASM_${#submitted[@]} \
    -v SINGULARITY_TMPDIR=${SINGULARITY_TMPDIR} \
    -v SINGULARITY_CACHEDIR=${SINGULARITY_CACHEDIR} \
    -v NXF_SINGULARITY_CACHEDIR=${SINGULARITY_CACHEDIR} \
    ${LAB_HOME}/workflows/wf-paired-end-illumina-assembly/_run_assembly.uge-nextflow

elif [[ ${#submitted[@]} -ge 1 ]] && \
  [[ ${HOSTNAME%%.*} == 'rosalind01' ]] || \
  [[ ${NODE_NUM} -ge 231 ]]; then
  # Submit Nextflow pipeline to Rosalind HPC
  qsub \
    -m ba \
    -q all.q \
    -v IN=${IN} \
    -v OUT=${OUT} \
    -o ${OUT}/pipeline_info \
    -e ${OUT}/pipeline_info \
    -M ${USER}@cdc.gov \
    -v LAB_HOME=${LAB_HOME} \
    -N ASM_${#submitted[@]} \
    -l max_runtime=72:00:00 \
    -v SINGULARITY_TMPDIR=${SINGULARITY_TMPDIR} \
    -v SINGULARITY_CACHEDIR=${SINGULARITY_CACHEDIR} \
    -v NXF_SINGULARITY_CACHEDIR=${SINGULARITY_CACHEDIR} \
    ${LAB_HOME}/workflows/wf-paired-end-illumina-assembly/_run_assembly.uge-nextflow

else
  echo -e "${RED_TXT}Biolinux/Aspen/Rosalind HPC is not detected.\nSubmission cancelled. ${COLOR_OFF}"
  exit 1
fi

# Display basename of FastQ files submitted
echo -e "${GREEN_TXT}\nInput directory:${COLOR_OFF} ${IN}"
echo -e "${GREEN_TXT}Output directory:${COLOR_OFF} ${OUT}"
echo -e "${GREEN_TXT}\nGenerating annotated assemblies for:"
for elem in "${submitted[@]}"; do
  echo "${elem}"
done | xargs -L3 | column -t
echo -e "${COLOR_OFF}"