run-pgap

#! /bin/bash

set -e

BACKEND=singularity

# ------------------------------------------------------------------------

OUT_DIR=pgap_out

function usage {
    echo 1>&2 "Usage: $0 [options] genome1.fna [...] [-- pgap.py args]"

    echo 1>&2 "-O ORGANISM_NAME"
    echo 1>&2 "-S STRAIN_NAME (required)"
    echo 1>&2 "-U - update PGAP, then exit"
    echo 1>&2 "-h - this message"
    echo 1>&2 "-k - keep temporary files"
    echo 1>&2 "-o OUT_DIR - Output directory [$OUT_DIR]"
    echo 1>&2 "-p PGAP_INPUT_DIR - Directory containing PGAP. Will download if not specified."
    echo 1>&2 "-s SUBMOL_YAML - submol.yaml to use instead of PGAP generated"
    echo 1>&2 "-u - update PGAP before running"

    echo 1>&2 ''
    echo 1>&2 "Exactly one of '-O', '-U', or '-s' option must be specified"

    exit "$@"
}    

while getopts 'O:S:V:Uhko:p:s:u' opt ; do
    case "$opt" in

	O) ORGANISM_NAME="$OPTARG" ;;
	S) STRAIN_NAME="$OPTARG" ;;
	U) UPDATE_PGAP=1 ; THEN_EXIT=1 ;;
	h) DISPLAY_HELP=1 ;;
	k) KEEP_FILES=1 ;;
	o) OUT_DIR="$OPTARG" ;;
	p) PGAP_INPUT_DIR="$OPTARG" ;;
	s) SUBMOL_YAML="$OPTARG" ;;
	u) UPDATE_PGAP=1 ;;

	\?) usage 1 ;;
	*) echo "Can't happen" ; exit 1 ;;
    esac
done
shift $((OPTIND-1))

if [ "$DISPLAY_HELP" ] ; then
    usage
fi

ARG_COUNT=
if [ "$SUBMOD_YAML" ] ; then ARG_COUNT+=1 ; fi
if [ "$ORGANISM_NAME" ] ; then ARG_COUNT+=1 ; fi
if [ "$THEN_EXIT" ] ; then ARG_COUNT+=1 ; fi
if [ "$ARG_COUNT" != 1 ] ; then
    echo 1>&2 "Exactly one of '-O', '-U', or '-s' option must be specified"    
    exit 1
fi

if [ -z "$THEN_EXIT" ] ; then
    if [ -z "$1" -o x"$1"x = x--x ] ; then
	echo 1>&2 "Missing genome file."
	exit 1
    fi
fi

# ------------------------------------------------------------------------

function run {
    echo 1>&2 + "$@"
    "$@"
}

if [ "$PGAP_INPUT_DIR" ] ; then
    if [ ! -e "$PGAP_INPUT_DIR" ] ; then
	echo 1>&2 "$PGAP_INPUT_DIR does not exist"
	exit 1
    fi
    export PGAP_INPUT_DIR
elif [ -d $HOME/scratch/pgap ] ; then
    export PGAP_INPUT_DIR=$HOME/scratch/pgap
else
    export PGAP_INPUT_DIR=${OUT_DIR}/pgap_dist
    echo "# Downloading PGAP to ${PGAP_INPUT_DIR}"
    (
	mkdir -p ${PGAP_INPUT_DIR}
	run cd ${PGAP_INPUT_DIR}
	run wget -q https://github.com/ncbi/pgap/raw/prod/scripts/pgap.py
	run chmod +x pgap.py
	run ./pgap.py -D ${BACKEND} --update --taxcheck
    )
    UPDATE_PGAP= # skip updating again
fi

# ------------------------------------------------------------------------

if [ "$UPDATE_PGAP" ] ; then
    (
	run	cd ${PGAP_INPUT_DIR}
	run rm -f pgap.py
	run wget -q https://github.com/ncbi/pgap/raw/prod/scripts/pgap.py
	run chmod +x pgap.py
	run ./pgap.py -D ${BACKEND} --update --taxcheck
    )
fi
if [ "$THEN_EXIT" ] ; then
    exit
fi

# ------------------------------------------------------------------------

TEMP_PATH=$(mktemp -d ${PGAP_INPUT_DIR}/tmp.XXXXXXXX)
TEMP_DIR=$(basename $TEMP_PATH)

mkdir -p ${TEMP_PATH}/input

while [ -n "$1" -a x"$1"x != x--x ] ; do
    cat "$1" >> ${TEMP_PATH}/input/genome.fasta
    shift 1
done
if [ x"$1"x == x--x ] ; then
    shift 1
fi

if [ "$SUBMOL_YAML" ] ; then
    cp --archive "$SUBMOL_YAML" ${TEMP_PATH}/input/submol.yaml

else
    cat <<EOF > ${TEMP_PATH}/input/submol.yaml 
organism:
    genus_species: ${ORGANISM_NAME}
EOF

fi

cat <<EOF > ${TEMP_PATH}/input/input.yaml
fasta:
  class: File
  location: genome.fasta
submol:
  class: File
  location: submol.yaml

EOF

# ------------------------------------------------------------------------

(
    run	cd ${PGAP_INPUT_DIR}
    ulimit -n 8192
    (
	set +e # keep going even if pgap.py fails
	run ./pgap.py -D ${BACKEND} --output ${TEMP_DIR}/output "$@" \
	    --report-usage-false --no-internet --taxcheck \
	    ${TEMP_DIR}/input/input.yaml
	true # keep going even if pgap.py fails
    )
)

# ------------------------------------------------------------------------

if [ ! -e ${TEMP_PATH}/output/annot.gbk ] ; then
    echo 2>&1 '# ***********************'
    echo 2>&1 '# * !!! PGAP Failed !!! *'
    echo 2>&1 '# ***********************'
    echo 2>&1 '#'
    echo 2>&1 '# Output can be found:' ${TEMP_PATH}/output
    if [ -e ${TEMP_PATH}/output/calls.tab ] ; then
	echo 2>&1 '# Head of:' ${TEMP_PATH}/output/calls.tab
	head ${TEMP_PATH}/output/calls.tab
    fi
    exit 1
fi

# ------------------------------------------------------------------------

mkdir -p ${OUT_DIR}
cp --archive ${TEMP_PATH}/output/* ${OUT_DIR}
if [ "$SUBMOL_YAML" ] ; then
    cp --archive ${TEMP_PATH}/input/*.yaml ${OUT_DIR}
fi

if [ -z "$KEEP_FILES" ] ; then
    (
	set +e
	# set -x
	chmod -R u+w ${TEMP_PATH}
	rm -rf ${TEMP_PATH}
    )
fi

# ------------------------------------------------------------------------