updated check for trailing tabs in motiffiles

bin/1.2_filter_motifs/compareBed.sh

@@ -8,17 +8,19 @@
 #             BED-format. 
 
 #             For parameter description, run the script without parameters or -h.
+#             The output is a file with the filtered footprints and the log file FilterMotivs.stats
 
-# One R script is used, called compareBed_runinfo.R, stored in the same directory
+# One R script is used, compareBed_runinfo.R, stored in the same directory.
 
 # default parameters
 workdir=$PWD
 output="newMotifs.bed"
 min=10
 max=200
-path=`echo $0 | sed 's/\/[^\/]*$/\//g'`
 help=false
 
+path=`echo $0 | sed 's/\/[^\/]*$/\//g'`
+
 # display help when no parameters chosen
 if [ $# -eq 0 ]
 then
@@ -32,7 +34,7 @@ do
 	key="$1"
 	if [[ ${2:0:1} == "-" ]]
 	then
-		echo "Each parameter needs a value (except the help parameter), values must not start with a '-'!"
+		echo "ERROR: Each parameter needs a value (except the help parameter), values must not start with a '-'!"
 		exit 1
 	fi
 
@@ -72,11 +74,6 @@ do
 			shift
 			shift
 		;;
-		-p|--path)
-			path="$2"
-			shift
-			shift
-		;;
 		-h|--help)
 			help=true
 			shift
@@ -94,7 +91,7 @@ if [ $count -gt 0 ]
 then
 	for i in ${wrong[@]}
 	do
-		echo wrong parameter $i
+		echo ERROR: wrong parameter $i
 	done
 
 	echo call script without parameters for help or call --help
@@ -104,11 +101,11 @@ fi
 # the help message
 if [ $help == true ]
 then
-	echo "This script utilies bedtools to select new footprints from data."
+	echo "This script utilizes bedtools to select new footprints from data."
 	echo "Therefore the data is compared with known footprint motifs."
-	echo "The output is a new .bed file with added sequence information and a column with flags for better sequences (1)"
-	echo "Required arguments are data and motifs, both in bed-format, and the fasta genome sequences."
-	echo "If a parameter is chosen, a value must be provided or an error will occur."
+	echo "The output is a new BED-file with added sequence information and a flag contains_maxpos (1/0)"
+	echo "Required arguments are data, motifs, both in bed-format and the fasta genome sequences."
+	echo "If a parameter is chosen, a value must be provided aswell. (exception -h)"
 	echo "--------------------"
 	echo "usage: compareBed.sh --data <path/to/file.bed> --motifs <path/to/known/motifs.bed> --fasta <path/to/genome.fasta> [optional_parameter <value>] ..."
 	echo "--------------------"
@@ -125,8 +122,7 @@ then
 	echo "              -max --max          maximum size of footprints; default is 80"
 	echo "              -o   --output       output path/file ; default dir current directory and filename is newMotifs.bed and newMotifs.bed.fasta"
 	echo "              -h --help           shows this help message"
-	echo "              -p --path           the path where the required script compareBed_runinfo.R is stored. Default: same path as this scripts path"
-exit 0
+	exit 0
 fi
 
 # summary of parameters
@@ -139,7 +135,6 @@ echo fasta: $fasta \(required\)
 echo min: $min
 echo max: $max
 echo output: $output
-echo relative path of subscript: $path
 
 # check required parameters
 if [ -z $data ] || [ -z $motifs ] || [ -z $fasta ]
@@ -149,10 +144,38 @@ then
 	echo required are: --data \<path/data.bed\> --motifs \<path/motifs.bed\> --fasta \<path/file.fasta\>
 	exit 1
 fi
+if [ ! -f $data ]
+then
+	echo ERROR
+	echo $data does not exist. Please check input parameter -d / --data
+	exit 1
+fi
+if [ ! -f $fasta ]
+then
+	echo ERROR
+	echo $fasta does not exist. Please check input parameter -f / --fasta
+	exit 1
+fi
+#check other parameters
+if [ $min -lt 0 ]
+then
+	min=10
+	echo "min can't be negative. Default value of 10 is choosen"
+fi
+if [ $max -lt $min ]
+then
+	max=200
+	echo "max must be greater than min. Default value of 200 is choosen"
+fi
+if [ ! -d $workdir ]
+then
+	echo ERROR
+	echo "directory $workdir does not exist. Please check parameter -w / --workdir"
+	exit 1
+fi
 
 # remove trailing tabs in footprints
 cat $data | sed 's/[ \t]*$//' > "$workdir"/filtered.bed
-all_empty=true
 
 # motiffiles either from a directory OR comma separated list
 if [ -d "$motifs" ]
@@ -166,6 +189,7 @@ else
 fi
 
 # check if files exist and if they are all empty (exiting if all empty)
+all_empty=true
 for i in ${motiffiles[@]}
 do
 	if [ -f $i ]
@@ -180,7 +204,7 @@ do
 			fi
 		fi
 	else
-		echo file $i does not exist
+		echo ERROR: file $i does not exist
 		echo please use correct paths. exiting.
 		exit 1
 	fi
@@ -202,18 +226,26 @@ temp_switch=true
 counter=1
 for i in ${motiffiles[@]}
 do
-	# remove trailing tabs in motiffile
-	sed -i 's/[ \t]*$//' $i
+	echo "$i  ---  $counter of ${#motiffiles[@]}"
+	# remove trailing tabs in motiffile, but only if the second line in the file ends with a tab.
+	# checking all lines for trailing tabs would be time consuming.
+	secnd_line=`sed -n 2p $i | tr '\t' '#'`
+	echo $secnd_line
+	if [[ ${secnd_line: -1} == "#" ]]
+	then
+		echo trailing tabs have been found. removing trailing tabs.
+		sed -i 's/[ \t]*$//' $i
+	fi
 
+	# bedtools comparisons
 	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
+		bedtools subtract -a "$workdir"/filtered_temp.bed -b $i > "$workdir"/filtered.bed
 	fi
-	echo "$i  ---  $counter of ${#motiffiles[@]}"
 	counter=`expr $counter + 1`
 done
 
@@ -253,5 +285,6 @@ else
 fi
 
 # add fasta sequences to bed and create fasta file
+echo getting sequences from fasta-file
 bedtools getfasta -fi $fasta -bed "$workdir"/filtered_flagged.bed -bedOut >> $output
 bedtools getfasta -name -fi $fasta -bed "$output" -fo "$output".fasta