parseScores

#!/bin/bash

# Parses a bunch of motifSearches (tipically from a specific method)
# Uses getScores and getThreshold
# A file named paramsTable.txt should be present, containing the Imin and I
# parameters for each regulator, as computed by motifParams
# Only those regulators having score > Imin are used to compute the score
# threshold.
# Some methods have p-values whose log10 score is used
#
# tag reg  n Imin          Ireg
# new ChvI 1 22.6739400442 14.6603348629
# new FixJ 1 22.6739400442 24.592962035
# new FixJ 2 22.6739400442 13.1111175173
# new FixK 1 22.6739400442 18.1515146439
#
# Outputs the same files with just the lines above threshold

OPTIND=1

log=0

while getopts "h?L" opt; do
    case "$opt" in
    h|\?)
        echo 'USAGE: parseScores IN_DIR OUT_DIR'
        echo '-L use the log10 score'
        exit 0
        ;;
    L)  log=1
        ;;
    esac
done

n=1                                                                                       
while [ $# -gt 0 ]; do                                                                    
        if [ $n -lt $OPTIND ]; then  
		# remove (shift) option arguments
		# until they are all gone                                                     
                let n=$n+1                                                                
                shift                                                                     
        else                                                                              
                break;                                                                    
        fi                                                                                
done

if [ "$#" -lt 2 ]; then
    echo 'USAGE: parseScores IN_DIR OUT_DIR'
    echo '-L use the log10 score'
    exit 0
fi

IN_DIR=$1
OUT_DIR=$2

ABOVE=""
BELOW=""

for file in $(ls $IN_DIR);                                                 
do
    file1=$(echo $file | sed 's/SM_b/SMb/g')
    tag=$(echo $file1 | awk 'BEGIN {FS="/"};{print $2}' | awk 'BEGIN {FS="_"};{print $1}');
    reg=$(echo $file1 | awk 'BEGIN {FS="_"};{print $4}');
    reg=$(echo $reg | sed 's/SMb/SM_b/g')
    count=$(echo $file1 | awk 'BEGIN {FS="_"};{print $5}' | awk 'BEGIN {FS="."};{print $1}');
    motifparam=$(cat paramsTable.txt | grep "$tag $reg $count");
    I_min=$(echo $motifparam | awk '{print $4}');
    I_reg=$(echo $motifparam | awk '{print $5}');

    compare=$(echo "$I_reg>$I_min" | bc -l); # bc -l permette di comparare due numeri con la virgla

    if [ $compare -gt 0 ]
    then
        ABOVE+=$IN_DIR/$file
        ABOVE+=" "
    else
        BELOW+=$IN_DIR/$file
        BELOW+=" "
    fi
done

if [ $log == 1 ]
then
    Threshold=$(./getScores $ABOVE | ./getThreshold -L);
else
    Threshold=$(./getScores $ABOVE | ./getThreshold);
fi
echo "Score threshold: $Threshold"

for file in $(ls $IN_DIR);	
do	
	file_1=$(echo $file | awk 'BEGIN {FS="/"};{print $NF}');
	if [ $log == 1 ]
    then
	    cat $IN_DIR/$file | awk -v t=$Threshold 'BEGIN {FS="\t"};{if($11<=t) print $0}' > $OUT_DIR/$file_1
	else
	    cat $IN_DIR/$file | awk -v t=$Threshold 'BEGIN {FS="\t"};{if($11>=t) print $0}' > $OUT_DIR/$file_1
	fi
done