run.sh

#!/bin/bash
set -ei

# This function only changes echo headers
# for user's sake.
function message() {
  # Define local variables
  local status=${1}         # Either INFO, CMD, ERROR or DOC
  local message="${2:-1}"   # Your message

  # Classic switch based on status
  if [ ${status} = INFO ]; then
    echo -e "\033[1;36m@INFO:\033[0m ${message}"
  elif [ ${status} = WARNING ]; then
    echo -e "\033[1;33m@WARNING:\033[0m ${message}"
  elif [ ${status} = ERROR ]; then
    echo -e "\033[41m@ERROR:\033[0m ${message}"
  elif [ ${status} = DOC ]; then
    echo -e "\033[0;33m@DOC:\033[0m ${message}"
  else
    error_handling ${LINENO} 1 "Unknown message type"
  fi
}

# This function will take error messages and exit the program
function error_handling() {
  # Geathering input parameter (message, third parameter is optionnal)
  echo -ne "\n"
  local parent_lineno="$1"
  local code="$2"
  local message="${3:-1}"

  # Checking the presence or absence of message
  if [[ -n "$message" ]] ; then
    # Case message is present
    message ERROR "Error on or near line ${parent_lineno}:\n ${message}"
    message ERROR "Exiting with status ${code}"
  else
    # Case message is not present
    message ERROR "Error on or near line ${parent_lineno}"
    message ERROR "Exiting with status ${code}"
  fi

  # Exiting with given error code
  exit "${code}"
}

function help_message() {
  message DOC "Hi, I'm functionnal only at IGR's Flamingo. Do not try to run "
  message DOC "me elsewhere. I won't work."
  echo ""
  message DOC "Thanks for using me as your script for running "
  message DOC "vcf-annotate-snpeff-snpsift. I'm very proud to be your script "
  message DOC "today, and I hope you'll enjoy working with me."
  echo ""
  message DOC "Every time you'll see a line starting with '@', "
  message DOC "it will be because I speak."
  message DOC "In fact, I always start my speech with :"
  message DOC "'\033[0;33m@DOC:\033[0m' when i't about my functions,"
  message DOC "'\033[1;36m@INFO:\033[0m' when it's about my intentions, "
  message DOC "'\033[1;33m@WARNING:\033[0m' for non-critical issues, "
  message DOC "'\033[41m@ERROR:\033[0m', I tell you when things go wrong."
  echo ""
  message DOC "I understand very fiew things, and here they are:"
  message DOC "-h | --help        Print this help message, then exit."
  message DOC "--vcf-dir          Path to VCF files (default: .)"
  message DOC "--hg19             Use GRCh37 annotations, instead of GRCh38.86"
  message DOC "--mm10             Use GRCm38 annotations, instead of GRCh38.86"
  message DOC "--mm9              Use GRCm37 annotations, instead of GRCh38.86"
  message DOC "Otherwise, run me without any arguments and I'll do magic."
  echo ""
  message DOC "A typical command line would be:"
  message DOC "bash /path/to/run.sh"
  exit 0
}

function export_hg37 {
  message INFO "Using HG19 annotations"
  declare -x GWASCAT_PATH="/mnt/beegfs/database/bioinfo/Index_DB/test.gwascat.tsv"
  declare -x DBNSFP_PATH="/mnt/beegfs/database/bioinfo/Index_DB/test.dbNSFP.vcf.gz"
  declare -x DBSNP_PATH="/mnt/beegfs/database/bioinfo/Index_DB/dbsnp.vcf"
  declare -x GENESETS_PATH="/mnt/beegfs/database/bioinfo/Index_DB/test.gmt"
  declare -x COSMIC_PATH="/mnt/beegfs/database/bioinfo/Index_DB/Cosmic.vcf"
  declare -x KAVIAR_PATH="/mnt/beegfs/database/bioinfo/Index_DB/kaviar.vcf"
  declare -x CLINVAR_PATH="/mnt/beegfs/database/bioinfo/Index_DB/clinvar.vcf"
  declare -x KGENOME_PATH="/mnt/beegfs/database/bioinfo/Index_DB/1000g.vcf"
  declare -x SNPEFF_DB_PATH="/mnt/beegfs/database/bioinfo/Index_DB/snpeff/data/GRCtest/"
  export GWASCAT_PATH DBNSFP_PATH DBSNP_PATH GENESETS_PATH COSMIC_PATH KAVIAR_PATH CLINVAR_PATH KGENOME_PATH SNPEFF_DB_PATH
}


function export_mm10 {
  message INFO "Using MM10 annotations"
  declare -x GWASCAT_PATH="/mnt/beegfs/database/bioinfo/Index_DB/test.gwascat.tsv"
  declare -x DBNSFP_PATH="/mnt/beegfs/database/bioinfo/Index_DB/test.dbNSFP.vcf.gz"
  declare -x DBSNP_PATH="/mnt/beegfs/database/bioinfo/Index_DB/dbsnp.vcf"
  declare -x GENESETS_PATH="/mnt/beegfs/database/bioinfo/Index_DB/test.gmt"
  declare -x COSMIC_PATH="/mnt/beegfs/database/bioinfo/Index_DB/Cosmic.vcf"
  declare -x KAVIAR_PATH="/mnt/beegfs/database/bioinfo/Index_DB/kaviar.vcf"
  declare -x CLINVAR_PATH="/mnt/beegfs/database/bioinfo/Index_DB/clinvar.vcf"
  declare -x KGENOME_PATH="/mnt/beegfs/database/bioinfo/Index_DB/1000g.vcf"
  declare -x SNPEFF_DB_PATH="/mnt/beegfs/database/bioinfo/Index_DB/snpeff/data/GRCtest/"
  export GWASCAT_PATH DBNSFP_PATH DBSNP_PATH GENESETS_PATH COSMIC_PATH KAVIAR_PATH CLINVAR_PATH KGENOME_PATH SNPEFF_DB_PATH
}


function export_mm9 {
  message INFO "Using MM9 annotations"
  declare -x GWASCAT_PATH="/mnt/beegfs/database/bioinfo/Index_DB/test.gwascat.tsv"
  declare -x DBNSFP_PATH="/mnt/beegfs/database/bioinfo/Index_DB/test.dbNSFP.vcf.gz"
  declare -x DBSNP_PATH="/mnt/beegfs/database/bioinfo/Index_DB/dbsnp.vcf"
  declare -x GENESETS_PATH="/mnt/beegfs/database/bioinfo/Index_DB/test.gmt"
  declare -x COSMIC_PATH="/mnt/beegfs/database/bioinfo/Index_DB/Cosmic.vcf"
  declare -x KAVIAR_PATH="/mnt/beegfs/database/bioinfo/Index_DB/kaviar.vcf"
  declare -x CLINVAR_PATH="/mnt/beegfs/database/bioinfo/Index_DB/clinvar.vcf"
  declare -x KGENOME_PATH="/mnt/beegfs/database/bioinfo/Index_DB/1000g.vcf"
  declare -x SNPEFF_DB_PATH="/mnt/beegfs/database/bioinfo/Index_DB/snpeff/data/GRCtest/"
  export GWASCAT_PATH DBNSFP_PATH DBSNP_PATH GENESETS_PATH COSMIC_PATH KAVIAR_PATH CLINVAR_PATH KGENOME_PATH SNPEFF_DB_PATH
}

VCF_DIR="${PWD}"
ANNOT="hg38"


while [[ $# -gt 0 ]]; do
  case "${1}" in
    -h|--help)
    help_message
    ;;
    --vcf-dir)
    VCF_DIR="${2}"
    shift 2
    ;;
    --hg19)
    ANNOT="hg19"
    shift
    ;;
    --mm10)
    ANNOT="mm10"
    shift
    ;;
    --mm9)
    ANNOT="mm9"
    shift
    ;;
    *)
    error_handling ${LINENO} 1 "Unknown argument ${1}"
    shift
    ;;
  esac
done


if [ ANNOT = hg19 ]; then
  export_hg19
elif [ ANNOT = mm10 ]; then
  export_mm10
elif [ ANNOT = mm9 ]; then
  export_mm9
else
  message INFO "Using HG38 annotations"
fi


CONDA='conda'
CONDA_VERSION="$(conda --version)"
CONDA_ENV="/mnt/beegfs/pipelines/rna-count-salmon/env"
[ "${CONDA_VERSION:?}" = "conda 4.9.2" ] || message WARNING "Your version of conda might not be up to date. Trying anyway with the rest of the pipeline."
which mamba > /dev/null 2>&1 && CONDA="mamba" || message WARNING "Mamba not found, falling back to conda."

# Loading conda
message INFO "Sourcing conda for users who did not source it before."
source "$(conda info --base)/etc/profile.d/conda.sh" && conda activate || exit error_handling "${LINENO}" 1 "Could not source conda environment."

# Check on environment variables: if env are missing
message INFO "Loading 'vcf-annotate-snpeff-snpsift' environment"
conda activate ${CONDA_ENV} || error_handling "${LINENO}" 2 "Could not activate the environment for vcf-annotate-snpeff-snpsift."

# then installation process did not work properly
message INFO "Running pipeline if and only if it is possible"
$(export -p | grep "ANNOT_LAUNCHER" --quiet) && python3 ${ANNOT_LAUNCHER:?} flamingo --vcf_dir ${VCF_DIR} || error_handling ${LINENO} 3 "Error while running: ${ANNOT_LAUNCHER}"