ftdmp-plot-jury-scores

#!/bin/bash

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

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

Options:
    --input-scoring-table     string  *  input scoring table file path
    --prefix-for-jury-scores  string     input column names prefix, default is 'RJS' \
    --output                  string  *  output plot file path
    --help | -h                          flag to display help message and exit

Example:
    ftdmp-plot-jury-scores --input-scoring-table ./scores/scoring_results.txt \
      --prefix-for-jury-scores 'raw_RJS'  --output ./plot.png

EOF
exit 1
}

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

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

FROM_TABLE=""
PREFIX_FOR_JURY_SCORES="RJS_"
OUTFILE=""
HELP_MODE="false"

while [[ $# > 0 ]]
do
	OPTION="$1"
	OPTARG="$2"
	shift
	case $OPTION in
	--input-scoring-table)
		FROM_TABLE="$OPTARG"
		shift
		;;
	--prefix-for-jury-scores)
		PREFIX_FOR_JURY_SCORES="$OPTARG"
		shift
		;;
	--output)
		OUTFILE="$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 "$FROM_TABLE" ]
then
	echo >&2 "Error: input scoring table file not specified"
	exit 1
fi

if [ ! -s "$FROM_TABLE" ]
then
	echo >&2 "Error: invalid input scoring table file '$FROM_TABLE'"
	exit 1
fi

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

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

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

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

cp "$FROM_TABLE" "$TMPLDIR/input.txt"

cd "$TMPLDIR"

cat "./input.txt" \
| head -1 \
| sed 's/^\s\+//' \
| sed 's/\s\+/ /g' \
| tr ' ' '\n' \
| egrep "^${PREFIX_FOR_JURY_SCORES}_top" \
| sed 's/^\(\S\+_top\)\(\S\+\)/\1\2 \2/' \
> "./top_columns_info.txt"

R --vanilla --args "$PREFIX_FOR_JURY_SCORES" > /dev/null << 'EOF'
args=commandArgs(TRUE);
prefix_for_jury=args[1];

dt=read.table("input.txt", header=TRUE, stringsAsFactors=FALSE);
maxcol=paste0(prefix_for_jury, "_max");

topcols_info=read.table("top_columns_info.txt", header=FALSE, stringsAsFactors=FALSE);
top_names=topcols_info$V1;
top_nums=topcols_info$V2;
if(top_nums[length(top_nums)]>=9999){top_nums[length(top_nums)]=top_nums[length(top_nums)-1]+5*(top_nums[length(top_nums)-1]-top_nums[length(top_nums)-2]);}

png("result.png", width=1000, height=600, units="px", pointsize=14);

dt=dt[order(dt[,maxcol]),];

plot(c(0, max(top_nums)), c(0, 1), type="n", xlab="top", ylab="ranks jury score", main=paste0("Jury scores, red ID = '", dt$ID[nrow(dt)], "'"));

for(i in 1:nrow(dt))
{
	sdt=data.frame(x=top_nums, y=as.numeric(dt[i, top_names]));
	sdt=sdt[which(sdt$y>0),];
	thecol=ifelse(i==nrow(dt), "#FF0000", "#777777");
	thelwd=ifelse(i==nrow(dt), 1.5, 1);
	points(sdt$x, sdt$y, type="l", col=thecol, lwd=thelwd);
	points(x=0, y=dt[i, maxcol], col=thecol);
}

dev.off();

EOF

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

cd - &> /dev/null

mkdir -p "$(dirname ${OUTFILE})"

mv "${TMPLDIR}/result.png" "$OUTFILE"