ftdmp-dock-two-monomers-with-hex

#!/bin/bash

function print_help_and_exit
{
cat >&2 << 'EOF'

'ftdmp-dock-two-monomers-with-hex' produces many dimeric structures by docking two monomeric structures

Options:
    --monomer1                string  *  static monomer input file path
    --monomer2                string  *  mobile monomer input file path
    --job-name                string     job name to use as prefix for identifiers
    --hex-macro-mode          string     flag to enable HEX macro mode, default is 'true'
    --hex-max-solutions       number     max number of docking solutions for HEX, default is 10000
    --hex-script              string     semicolon-sparated additional commands for HEX, default is ''
    --hex-swap-and-repeat     string     flag to run HEX twice with monomers swapped, default is 'false'
    --sbatch-parameters       string     sbatch parameters to run on cluster, default is ''
    --help | -h                          flag to display help message and exit

Standard output:
    space-separated table of docking results
    
Examples:

    ftdmp-dock-two-monomers-with-hex --monomer1 monomer1.pdb --monomer2 monomer2.pdb --hex-max-solutions 30000

EOF
exit 1
}

readonly ZEROARG=$0
ALLARGS=("$@")

if [ -z "$1" ]
then
	print_help_and_exit
fi

if [ -z "$FTDMPDIR" ]
then
	export FTDMPDIR="$( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )"
	export PATH="${FTDMPDIR}/core/voronota/expansion_js:${FTDMPDIR}:${PATH}"
fi

command -v voronota-js &> /dev/null || { echo >&2 "Error: 'voronota-js' executable not in binaries path"; exit 1; }
command -v ftdmp-prepare-monomer &> /dev/null || { echo >&2 "Error: 'ftdmp-prepare-monomer' executable not in binaries path"; exit 1; }
command -v hex &> /dev/null || { echo >&2 "Error: 'hex' executable not in binaries path"; exit 1; }

MONOMERFILE1=""
MONOMERFILE2=""
JOBNAME=""
HEX_MACRO_MODE="true"
HEX_MAX_SOLUTIONS="10000"
HEX_SCRIPT=""
HEX_SWAP_AND_REPEAT="false"
SBATCH_PARAMETERS=""
HELP_MODE="false"

while [[ $# > 0 ]]
do
	OPTION="$1"
	OPTARG="$2"
	shift
	case $OPTION in
	--monomer1)
		MONOMERFILE1="$OPTARG"
		shift
		;;
	--monomer2)
		MONOMERFILE2="$OPTARG"
		shift
		;;
	--job-name)
		JOBNAME="$OPTARG"
		shift
		;;
	--hex-macro-mode)
		HEX_MACRO_MODE="$OPTARG"
		shift
		;;
	--hex-max-solutions)
		HEX_MAX_SOLUTIONS="$OPTARG"
		shift
		;;
	--hex-script)
		HEX_SCRIPT="$OPTARG"
		shift
		;;
	--hex-swap-and-repeat)
		HEX_SWAP_AND_REPEAT="$OPTARG"
		shift
		;;
	--sbatch-parameters)
		SBATCH_PARAMETERS="$OPTARG"
		shift
		;;
	-h|--help)
		HELP_MODE="true"
		;;
	*)
		echo >&2 "Error: invalid command line option '$OPTION'"
		exit 1
		;;
	esac
done

if [ "$HELP_MODE" == "true" ]
then
	print_help_and_exit
fi

if [ -z "$MONOMERFILE1" ]
then
	echo >&2 "Error: no static monomer input file specified"
	exit 1
fi

if [ ! -s "$MONOMERFILE1" ]
then
	echo >&2 "Error: no static monomer input file '$MONOMERFILE1'"
	exit 1
fi

if [ -z "$MONOMERFILE2" ]
then
	echo >&2 "Error: no mobile monomer input file specified"
	exit 1
fi

if [ ! -s "$MONOMERFILE2" ]
then
	echo >&2 "Error: no mobile monomer input file '$MONOMERFILE2'"
	exit 1
fi

if [ -z "$HEX_MAX_SOLUTIONS" ] || [ "$HEX_MAX_SOLUTIONS" -ne "$HEX_MAX_SOLUTIONS" ] || [ "$HEX_MAX_SOLUTIONS" -lt "101" ]
then
	echo >&2 "Error: invalid number of docking solutions '$HEX_MAX_SOLUTIONS', must be a positive number larger than 100"
	exit 1
fi

readonly TMPLDIR=$(mktemp -d)
trap "rm -r $TMPLDIR" EXIT

if [ -n "$SBATCH_PARAMETERS" ]
then
	mkdir -p "$TMPLDIR/slurm_logs"
	
	sbatch -o "$TMPLDIR/slurm_logs/slurmjob-%j.out" -e "$TMPLDIR/slurm_logs/slurmjob-%j.err" $SBATCH_PARAMETERS "$ZEROARG" "${ALLARGS[@]}" --sbatch-parameters "" \
	| egrep '^Submitted batch job ' \
	| awk '{print $4}' \
	> "$TMPLDIR/slurm_job_ids"
	
	sleep 1
	REMAINING_SLURM_JOBS="$(squeue | grep -f "$TMPLDIR/slurm_job_ids" | wc -l)"
	while [ "$REMAINING_SLURM_JOBS" -gt "0" ]
	do
		sleep 5
		REMAINING_SLURM_JOBS="$(squeue | grep -f "$TMPLDIR/slurm_job_ids" | wc -l)"
	done
	
	find "$TMPLDIR/slurm_logs/" -type f -name '*.err' -not -empty | xargs -L 1 cat >&2
	find "$TMPLDIR/slurm_logs/" -type f -name '*.out' -not -empty | xargs -L 1 cat
	exit 0
fi

ftdmp-prepare-monomer --input "$MONOMERFILE1" --output-for-ftdock "$TMPLDIR/monomer1.pdb"

if [ ! -s "$TMPLDIR/monomer1.pdb" ]
then
	echo >&2 "Error: failed to prepare monomer 1"
	exit 1
fi

ftdmp-prepare-monomer --input "$MONOMERFILE2" --output-for-ftdock "$TMPLDIR/monomer2.pdb"

if [ ! -s "$TMPLDIR/monomer2.pdb" ]
then
	echo >&2 "Error: failed to prepare monomer 2"
	exit 1
fi

cd "$TMPLDIR"

{
	if [ "$HEX_MACRO_MODE" == "true" ]
	then
		echo "activate_macro_model"
	fi
	
	echo "$HEX_SCRIPT"
	
	echo "max_docking_solutions ${HEX_MAX_SOLUTIONS}"
	
	echo "activate_docking"
} \
| tr ';' '\n' \
| sed 's/^\s\+//' \
| sed 's/\s\+$//' \
> ./common_settings.mac

{
	echo "open_receptor ${TMPLDIR}/monomer1.pdb"
	echo "open_ligand ${TMPLDIR}/monomer2.pdb"
	cat ./common_settings.mac
	echo "save_matrix ${TMPLDIR}/matrix_main.hex"
} \
> ./full_settings_main.mac

cat ./full_settings_main.mac | hex &> /dev/null

if [ ! -s "./matrix_main.hex" ]
then
	echo >&2 "Error: failed to dock with HEX"
	exit 1
fi

if [ "$HEX_SWAP_AND_REPEAT" == "true" ]
then
	{
		echo "open_receptor ${TMPLDIR}/monomer2.pdb"
		echo "open_ligand ${TMPLDIR}/monomer1.pdb"
		cat ./common_settings.mac
		echo "save_matrix ${TMPLDIR}/matrix_alt.hex"
	} \
	> ./full_settings_alt.mac

	cat ./full_settings_alt.mac | hex &> /dev/null
fi

{
	cat "./matrix_main.hex"
	if [ -s "./matrix_alt.hex" ]
	then
		cat "./matrix_alt.hex"
	fi
} \
| egrep '^LigandMatrix:' -A 3 \
| egrep -v '^--$' \
| sed 's/\s\+/ /g' \
| sed 's/^\s//' \
| sed 's/\s$//' \
| sed 's/^LigandMatrix//' \
| tr '\n' ' ' \
| tr ':' '\n' \
| egrep . \
| awk '{print sprintf("%g %g %g %g %g %g %g %g %g %g %g %g", $4, $8, $12, $1, $2, $3, $5, $6, $7, $9, $10, $11);}' \
> ./reformatted_matrices.txt

if [ ! -s "./reformatted_matrices.txt" ]
then
	echo >&2 "Error: failed to reformat matrices"
	exit 1
fi

{
echo "ID scscore scale x y z a1 a2 a3 a4 a5 a6 a7 a8 a9"

cat ./reformatted_matrices.txt \
| awk -v jobname="$JOBNAME" '{print jobname "hex" NR " 0 0 " $0}'
} \
> ./results.txt

cat ./results.txt