Jannik hamp patch 1

bin/1.2_filter_motifs/abs_max_score.R

@@ -0,0 +1,24 @@
+#!/bin/Rscript
+
+# author: Jannik Hamp
+# email: jannik.hamp@googlemail.com
+
+# desciption: This script gets called by compareBed.sh
+
+# It calculates the absolute position of maximum score of the footprints.
+# This information is later used for an additional flag column in the bed file.
+
+library(data.table)
+args = commandArgs(TRUE)
+file = args[1]
+
+tab = fread(file, sep='\t')
+
+# check if data has less than 9 columns
+if (ncol(tab) < 9) {
+	stop("footprint file has less than 9 columns. exiting.")
+}
+
+tab[[8]] = tab[[2]] + tab[[8]]
+
+fwrite(tab, file, row.names=FALSE, col.names=FALSE, quote=FALSE, sep='\t')

bin/1.2_filter_motifs/compareBed.sh

@@ -10,7 +10,7 @@
 #             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, compareBed_runinfo.R, stored in the same directory.
+# Two R scripts are used, compareBed_runinfo.R and abs_max_score.R, stored in the same directory.
 
 # default parameters
 workdir=$PWD
@@ -181,7 +181,7 @@ cat $data | sed 's/[ \t]*$//' > "$workdir"/filtered.bed
 if [ -d "$motifs" ]
 then
 	# 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"`)
+	declare -a motiffiles=(`ls $motifs | grep -i '.bed$' | sed "s|^|$motifs\/|g" | tr '\n' ' ' | sed "s|//|/|g"`)
 
 # the else case means, that the motiffiles were passed comma separated with no whitespace.
 else
@@ -219,6 +219,14 @@ if [ $all_empty == true ]
 		exit 1
 fi
 
+#calculate absolute max_score position
+Rscript abs_max_score.R "$workdir"/filtered.bed
+# check if Rscript executed without errors
+if [ $? -gt 0 ] 
+then
+	exit 1
+fi
+
 # comparing unknown footprints with regions of known motifs
 # comparison is done iteratively
 # remove overlapping regions in unknown footprints

bin/1.2_filter_motifs/compareBed_runinfo.R

@@ -8,7 +8,7 @@
 #             adds a column with a flag "contains_maxpos"
 #             and creates a file with information of the bedtool comparison
 
-# parameters: Parameters are not named. Respect the parameter order.
+# parameters: Parameters are not named, respect the parameter order.
 #             min:            minimum footprint size threshold 
 #             max:            maximum footprint size threshold
 #             input_raw:      unfiltered BED-file