added information for logfile

bin/1.2_filter_motifs/compareBed.sh

@@ -18,23 +18,19 @@ min=10
 max=200
 path=`echo $0 | sed 's/\/[^\/]*$/\//g'`
 help=false
-# required parameters given
-da=false
-mo=false
-fa=false
 
 # display help when no parameters chosen
 if [ $# -eq 0 ]
 then
-	he=true
+	help=true
 fi
 
 # parsing parameters
 wrong=()
 while [[ $# -gt 0 ]]
 do
 	key="$1"
-	if [[ "^-" =~ $2 ]]
+	if [[ ${2:0:1} == "-" ]]
 	then
 		echo "Each parameter needs a value (except the help parameter), values must not start with a '-'!"
 		exit 1
@@ -43,13 +39,11 @@ do
 	case $key in
 		-d|--data)
 			data="$2"
-			da=true
 			shift
 			shift
 		;;
 		-m|--motifs)
 			motifs="$2"
-			mo=true
 			shift
 			shift
 		;;
@@ -60,7 +54,6 @@ do
 		;;
 		-f|--fasta)
 			fasta="$2"
-			fa=true
 			shift
 			shift
 		;;
@@ -102,10 +95,10 @@ then
 	for i in ${wrong[@]}
 	do
 		echo wrong parameter $i
-		echo call script without parameters for help or call --help
-		echo exit
 	done
-exit 1
+
+	echo call script without parameters for help or call --help
+	exit 1
 fi
 
 # the help message
@@ -151,105 +144,78 @@ echo path of scripts: $path
 # check required parameters
 if [ -z $data ] || [ -z $motifs ] || [ -z $fasta ]
 then
+	echo ERROR
 	echo required parameters not given.
 	echo required are: --data \<path/data.bed\> --motifs \<path/motifs.bed\> --fasta \<path/file.fasta\>
 	exit 1
 fi
 
-# comparing unknown footprints with regions of known motifs
-# comparison is done iteratively
-# remove overlapping regions in unknown footprints
-
 # remove trailing tabs in footprints
 cat $data | sed 's/[ \t]*$//' > "$workdir"/filtered.bed
-temp_switch=true
 all_empty=true
+
+# motiffiles either from a directory OR comma separated list
 if [ -d "$motifs" ]
 then
-	# check if all motiffiles are empty/only consist of header. exit if all are empty 
-	for i in "$motifs"/*.bed
-	do
+	# creates an array of all files with bed in its name in the directory $motifs  
+	declare -a motiffiles=(`ls $motifs | grep bed | sed "s|^|$motifs\/|g" | tr '\n' ' ' | sed "s|//|/|g"`)
+
+# the else case means, that the motiffiles were passed comma separated with no whitespace.
+else
+	declare -a motiffiles=(`echo $motifs | sed 's/,/ /g'`)
+fi
+
+# check if files exist and if they are all empty (exiting if all empty)
+for i in ${motiffiles[@]}
+do
+	if [ -f $i ]
+	then
 		if [ $all_empty == true ]
 		then
 			lines=`cat $i | wc -l`
 			if [ $lines -gt 1 ]
 			then
 				all_empty=false
+				break
 			fi
 		fi
-	done
-	if [ $all_empty == true ]
-	then
-		echo All motiffiles were empty!
-		echo Fix motiffiles and try again.
-		echo exiting
+	else
+		echo file $i does not exist
+		echo please use correct paths. exiting.
 		exit 1
 	fi
+done
 
-	# bedtools comparisons
-	for i in "$motifs"/*.bed
-	do
-		# remove trailing tabs in motiffile
-		sed -i 's/[ \t]*$//' $i
-
-		if [ $temp_switch == true ]
-		then
-			temp_switch=false
-			bedtools subtract -a "$workdir"/filtered.bed -b $i > "$workdir"/filtered_temp.bed
-		else
-			temp_switch=true
-			bedtools subtract -a "$workdir"/filtered_temp.bed	-b $i > "$workdir"/filtered.bed
-		fi
-		echo $i
-	done
-
-# the else case means, that the motiffiles were passed comma separated with no whitespace.
-else
-	declare -a motiffiles=(`echo $motifs | sed 's/,/ /g'`)
-	# check if files exist and if they are all empty (exiting if all empty)
-	for i in ${motiffiles[@]}
-	do
-		if [ -f $i ]
-		then
-			if [ $all_empty == true ]
-			then
-				lines=`cat $i | wc -l`
-				if [ $lines -gt 1 ]
-				then
-					all_empty=false
-				fi
-			fi
-		else
-			echo file $i does not exist
-			echo please use correct paths. exiting.
-			exit 1
-		fi
-	done
-	if [ $all_empty == true ]
+# error report of rare case of only empty motiffiles
+if [ $all_empty == true ]
 	then
+		echo ERROR
 		echo All motiffiles were empty!
 		echo Fix motiffiles and try again.
-		echo exiting
 		exit 1
-	fi	
+fi
 
-	# bedtools comparisons
-	for i in ${motiffiles[@]}
-	do
-		# remove trailing tabs in motiffile
-		sed -i 's/[ \t]*$//' $i
+# comparing unknown footprints with regions of known motifs
+# comparison is done iteratively
+# remove overlapping regions in unknown footprints
+temp_switch=true
+counter=1
+for i in ${motiffiles[@]}
+do
+	# remove trailing tabs in motiffile
+	sed -i 's/[ \t]*$//' $i
 
-		if [ $temp_switch == true ]
-		then
-			help=false
-			bedtools subtract -a "$workdir"/filtered.bed -b $i > "$workdir"/filtered_temp.bed
-		else
-			temp_switch=true
-			bedtools subtract -a "$workdir"/filtered_temp.bed	-b $i > "$workdir"/filtered.bed
-		fi
-		echo $i
-	done
-fi
+	if [ $temp_switch == true ]
+	then
+		temp_switch=false
+		bedtools subtract -a "$workdir"/filtered.bed -b $i > "$workdir"/filtered_temp.bed
+	else
+		temp_switch=true
+		bedtools subtract -a "$workdir"/filtered_temp.bed	-b $i > "$workdir"/filtered.bed
+	fi
+	echo "$i  ---  $counter of ${#motiffiles[@]}"
+	counter=`expr $counter + 1`
+done
 
 # get file of last iteration an write its content into filtered.bed
 if [ $temp_switch == false ]
@@ -259,7 +225,7 @@ fi
 
 # remove short/long motivs, make unique ids (relevant for some splitted tfbs from subtract) and handle maxScorePosition
 # also creates a small output file with information about the comparison
-Rscript $path/compareBed_runinfo.R $min $max $data "$workdir"/filtered.bed "$workdir"/filtered_flagged.bed
+Rscript $path/compareBed_runinfo.R $min $max $data "$workdir"/filtered.bed "$workdir"/filtered_flagged.bed "$workdir"/FilterMotifs.stats
 # check if Rscript executed without errors
 if [ $? -gt 0 ] 
 then
@@ -271,9 +237,19 @@ first_line=`sed -n 1p $data | sed "s/$/\tcontains_maxpos\tsequence/"`
 if [[ ${first_line:0:1} == "#" ]]
 then
 	echo "$first_line" > $output
+	# add some final values to the log file
+	fp_initial=`cat $data | wc -l`
+	fp_initial=`expr $fp_initial - 1`
+	fp_final=`cat "$workdir"/filtered.bed | wc -l`
+	fp_final=`expr $fp_final - 1`
+	echo $fp_initial | sed 's/^/initial number of footprints: /g' >> "$workdir"/FilterMotifs.stats
+	echo $fp_final | sed 's/^/number of footprints after subtract: /g' >> "$workdir"/FilterMotifs.stats
 else
 	# output will be overwritten if it exists
 	rm -f $output
+	# add some final values to the log file
+	cat $data | wc -l | sed 's/^/initial number of footprints: /g' >> "$workdir"/FilterMotifs.stats
+	cat "$workdir"/filtered.bed | wc -l | sed 's/^/number of footprints after subtract: /g' >> "$workdir"/FilterMotifs.stats
 fi
 
 # add fasta sequences to bed and create fasta file