ftdmp-plot-jury-diagnostics

#!/bin/bash

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

'ftdmp-plot-jury-scores' plots jury scores from a scoring table

Options:
    --input-prefix            string  *  input file path prefix
    --output-prefix           string  *  output plot file path prefix
    --help | -h                          flag to display help message and exit

Example:
    ftdmp-plot-jury-diagnostics --input-prefix ./scores/scoring_diagnostic_ --output ./plot.png

EOF
exit 1
}

################################################################################

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

INPUT_PREFIX=""
OUTPUT_PREFIX=""
HELP_MODE="false"

while [[ $# > 0 ]]
do
	OPTION="$1"
	OPTARG="$2"
	shift
	case $OPTION in
	--input-prefix)
		INPUT_PREFIX="$OPTARG"
		shift
		;;
	--output-prefix)
		OUTPUT_PREFIX="$OPTARG"
		shift
		;;
	-h|--help)
		HELP_MODE="true"
		;;
	*)
		echo >&2 "Error: invalid command line option '$OPTION'"
		exit 1
		;;
	esac
done

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

if [ -z "$INPUT_PREFIX" ]
then
	echo >&2 "Error: input file path prefix not specified"
	exit 1
fi

if [ -z "$OUTPUT_PREFIX" ]
then
	echo >&2 "Error: output plot file prefix not specified"
	exit 1
fi

command -v R &> /dev/null || { echo >&2 "Error: 'R' executable not in binaries path"; exit 1; }

CLUSTER_COUNTS_FILE="${INPUT_PREFIX}cluster_counts_by_threshold.txt"

if [ ! -s "$CLUSTER_COUNTS_FILE" ]
then
	echo >&2 "Error: missing input file '$CLUSTER_COUNTS_FILE'"
	exit 1
fi

################################################################################

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

cp "$CLUSTER_COUNTS_FILE" "$TMPLDIR/cluster_counts.txt"

cd "$TMPLDIR"

R --vanilla > /dev/null << 'EOF'
dt=read.table("cluster_counts.txt", header=FALSE, stringsAsFactors=FALSE);
png("cluster_counts.png", width=1000, height=600, units="px", pointsize=14);
plot(dt$V1, dt$V2, xlab="threshold", ylab="clusters", main="Cluster counts vs thresholds");
dev.off();
EOF

if [ ! -s "${TMPLDIR}/cluster_counts.png" ]
then
	echo >&2 "Error: no output plot produced for '$CLUSTER_COUNTS_FILE'"
	exit 1
fi

cd - &> /dev/null

mkdir -p "$(dirname ${OUTPUT_PREFIX}suffix)"

mv "${TMPLDIR}/cluster_counts.png" "${OUTPUT_PREFIX}cluster_counts.png"