Mercurial > repos > davidvanzessen > mutation_analysis

diff wrapper.sh @ 110:ade5cf6fd2dc draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author davidvanzessen
date Tue, 02 Aug 2016 08:30:23 -0400
parents 6add3e66f4fa
children e7b550d52eb7
line wrap: on
line diff
--- a/wrapper.sh	Thu Jul 14 07:29:56 2016 -0400
+++ b/wrapper.sh	Tue Aug 02 08:30:23 2016 -0400
@@ -1,5 +1,5 @@
 #!/bin/bash
-set -e
+#set -e
 dir="$(cd "$(dirname "$0")" && pwd)"
 input=$1
 method=$2
@@ -17,13 +17,13 @@
 class_filter=${13}
 mkdir $outdir
 
+tar -xzf $dir/style.tar.gz -C $outdir
+
 echo "---------------- read parameters ----------------"
 echo "---------------- read parameters ----------------<br />" > $log
 
 echo "unpacking IMGT file"
 
-
-
 type="`file $input`"
 if [[ "$type" == *"Zip archive"* ]] ; then
 	echo "Zip archive"
@@ -67,7 +67,11 @@
 	sequence_index=$(cat $PWD/summary.txt | grep -o -P ".+\tSequence" | grep -o -P "\t" | wc -l)
 	sequence_index=$((sequence_index+1))
 	
-	cat $PWD/summary.txt | tail -n+2 | cut -f ${ID_index},${sequence_index} | awk '{print ">" $1 "\n" $2}' > $PWD/sequences.fasta
+	cat $PWD/summary.txt | tail -n+2 | cut -f ${ID_index},${sequence_index} | awk '{print ">" $1 "\n" $2}' > $PWD/sequences.tmp
+	
+	cat $PWD/sequences.tmp | grep -B1 -vE "^$" sequences.fasta #filter out empty sequences
+	
+	rm $PWD/sequences.tmp
 	
 	echo -e "qseqid\tsseqid\tpident\tlength\tmismatch\tgapopen\tqstart\tqend\tsstart\tsend\tevalue\tbitscore" > $outdir/identified_genes.txt
 	${BLASTN_DIR}/blastn -task blastn -db $dir/subclass_definition.db -query $PWD/sequences.fasta -outfmt 6 >> $outdir/identified_genes.txt
@@ -149,12 +153,15 @@
 echo "---------------- aa_histogram.r ----------------<br />" >> $log
 
 Rscript $dir/aa_histogram.r $outdir/aa_id_mutations.txt $outdir/absent_aa_id.txt "ca,cg,cm" $outdir/ 2>&1
-mv $outdir/aa_histogram_.png $outdir/aa_histogram.png
-mv $outdir/aa_histogram_.txt $outdir/aa_histogram.txt
+if [ -e "$outdir/aa_histogram_.png" ]; then
+        mv $outdir/aa_histogram_.png $outdir/aa_histogram.png
+        mv $outdir/aa_histogram_.txt $outdir/aa_histogram.txt
+fi
 
 genes=(ca ca1 ca2 cg cg1 cg2 cg3 cg4 cm)
 
 funcs=(sum mean median)
+funcs=(sum)
 
 echo "---------------- sequence_overview.r ----------------"
 echo "---------------- sequence_overview.r ----------------<br />" >> $log
@@ -172,6 +179,10 @@
 done < $outdir/sequence_overview/ntoverview.txt
 
 echo "<html><center><h1>$title</h1></center>" > $output
+echo "<script type='text/javascript' src='jquery-1.11.0.min.js'></script>" >> $output
+echo "<script type='text/javascript' src='tabber.js'></script>" >> $output
+echo "<script type='text/javascript' src='script.js'></script>" >> $output
+echo "<link rel='stylesheet' type='text/css' href='style.css'>" >> $output
 
 #display the matched/unmatched for clearity
 
@@ -188,6 +199,10 @@
 
 echo "---------------- main tables ----------------"
 echo "---------------- main tables ----------------<br />" >> $log
+
+echo "<div class='tabber'>" >> $output
+echo "<div class='tabbertab' title='SHM Overview'>" >> $output
+
 for func in ${funcs[@]}
 do
 
@@ -221,9 +236,84 @@
 	#echo "<a href='data_${func}.txt'>Download data</a>" >> $output
 done
 
-echo "---------------- download links ----------------"
-echo "---------------- download links ----------------<br />" >> $log
+echo "</div>" >> $output #SHM overview tab end
+
+echo "---------------- images ----------------"
+echo "---------------- images ----------------<br />" >> $log
+
+echo "<div class='tabbertab' title='SHM Frequency'>" >> $output
+
+if [ -a $outdir/scatter.png ]
+then
+	echo "<img src='scatter.png'/><br />" >> $output
+	echo "<a href='scatter.txt'>download data</a><br />" >> $output
+fi
+if [ -a $outdir/frequency_ranges.png ]
+then
+	echo "<img src='frequency_ranges.png'/><br />" >> $output
+	echo "<a href='frequency_ranges_classes.txt'>download class data</a><br />" >> $output
+	echo "<a href='frequency_ranges_subclasses.txt'>download subclass data</a><br />" >> $output
+fi
+
+echo "</div>" >> $output #SHM frequency tab end
+
+echo "<div class='tabbertab' title='Transition tables'>" >> $output
+
+for gene in ${genes[@]}
+do
+	echo "<table border='1'><caption>$gene transition table</caption>" >> $output
+	while IFS=, read from a c g t
+		do
+			echo "<tr><td>$from</td><td>$a</td><td>$c</td><td>$g</td><td>$t</td></tr>" >> $output
+	done < $outdir/transitions_${gene}_sum.txt
+	echo "</table>" >> $output
+done
 
+echo "<table border='1'><caption>All transition table</caption>" >> $output
+while IFS=, read from a c g t
+	do
+		echo "<tr><td>$from</td><td>$a</td><td>$c</td><td>$g</td><td>$t</td></tr>" >> $output
+done < $outdir/transitions_all_sum.txt
+echo "</table>" >> $output
+
+echo "</div>" >> $output #transition tables tab end
+
+echo "<div class='tabbertab' title='Antigen Selection'>" >> $output
+
+if [ -a $outdir/aa_histogram.png ]
+then
+	echo "<img src='aa_histogram.png'/><br />" >> $output
+	echo "<a href='aa_histogram.txt'>download data</a><br />" >> $output
+	echo "<img src='aa_histogram_ca.png'/><br />" >> $output
+	echo "<a href='aa_histogram_ca.txt'>download data</a><br />" >> $output
+	echo "<img src='aa_histogram_cg.png'/><br />" >> $output
+	echo "<a href='aa_histogram_cg.txt'>download data</a><br />" >> $output
+	echo "<img src='aa_histogram_cm.png'/><br />" >> $output
+	echo "<a href='aa_histogram_cm.txt'>download data</a><br />" >> $output
+fi
+
+echo "<embed src='baseline_ca.pdf' width='700px' height='1000px'>" >> $output
+echo "<embed src='baseline_cg.pdf' width='700px' height='1000px'>" >> $output
+echo "<embed src='baseline_cm.pdf' width='700px' height='1000px'>" >> $output
+
+echo "</div>" >> $output #antigen selection tab end
+
+echo "<div class='tabbertab' title='CSR'>" >> $output
+
+if [ -a $outdir/ca.png ] 
+then
+	echo "<img src='ca.png'/><br />" >> $output
+	echo "<a href='ca.txt'>download data</a><br />" >> $output
+fi
+if [ -a $outdir/cg.png ]
+then
+	echo "<img src='cg.png'/><br />" >> $output
+	echo "<a href='cg.txt'>download data</a><br />" >> $output
+fi
+
+echo "</div>" >> $output #CSR tab end
+
+echo "<div class='tabbertab' title='Downloads'>" >> $output
 
 echo "<a href='unmatched.txt'>unmatched</a><br />" >> $output
 echo "<a href='motif_per_seq.txt'>motif per sequence</a><br />" >> $output
@@ -250,61 +340,9 @@
 echo "<a href='new_IMGT_cg.txz'>Filtered cg IMGT zip</a><br />" >> $output
 echo "<a href='new_IMGT_cm.txz'>Filtered cm IMGT zip</a><br />" >> $output
 
-
-echo "---------------- images ----------------"
-echo "---------------- images ----------------<br />" >> $log
+echo "</div>" >> $output #downloads tab end
 
-echo "<img src='all.png'/><br />" >> $output
-echo "<a href='all.txt'>download data</a><br />" >> $output
-if [ -a $outdir/ca.png ] 
-then
-	echo "<img src='ca.png'/><br />" >> $output
-	echo "<a href='ca.txt'>download data</a><br />" >> $output
-fi
-if [ -a $outdir/cg.png ]
-then
-	echo "<img src='cg.png'/><br />" >> $output
-	echo "<a href='cg.txt'>download data</a><br />" >> $output
-fi
-if [ -a $outdir/scatter.png ]
-then
-	echo "<img src='scatter.png'/><br />" >> $output
-	echo "<a href='scatter.txt'>download data</a><br />" >> $output
-fi
-if [ -a $outdir/frequency_ranges.png ]
-then
-	echo "<img src='frequency_ranges.png'/><br />" >> $output
-	echo "<a href='frequency_ranges_classes.txt'>download class data</a><br />" >> $output
-	echo "<a href='frequency_ranges_subclasses.txt'>download subclass data</a><br />" >> $output
-fi
-if [ -a $outdir/aa_histogram.png ]
-then
-	echo "<img src='aa_histogram.png'/><br />" >> $output
-	echo "<a href='aa_histogram.txt'>download data</a><br />" >> $output
-	echo "<img src='aa_histogram_ca.png'/><br />" >> $output
-	echo "<a href='aa_histogram_ca.txt'>download data</a><br />" >> $output
-	echo "<img src='aa_histogram_cg.png'/><br />" >> $output
-	echo "<a href='aa_histogram_cg.txt'>download data</a><br />" >> $output
-	echo "<img src='aa_histogram_cm.png'/><br />" >> $output
-	echo "<a href='aa_histogram_cm.txt'>download data</a><br />" >> $output
-fi
-
-for gene in ${genes[@]}
-do
-	echo "<table border='1'><caption>$gene transition table</caption>" >> $output
-	while IFS=, read from a c g t
-		do
-			echo "<tr><td>$from</td><td>$a</td><td>$c</td><td>$g</td><td>$t</td></tr>" >> $output
-	done < $outdir/transitions_${gene}_sum.txt
-	echo "</table>" >> $output
-done
-
-echo "<table border='1'><caption>All transition table</caption>" >> $output
-while IFS=, read from a c g t
-	do
-		echo "<tr><td>$from</td><td>$a</td><td>$c</td><td>$g</td><td>$t</td></tr>" >> $output
-done < $outdir/transitions_all_sum.txt
-echo "</table>" >> $output
+echo "</div>" >> $output #tabs end 
 
 echo "</html>" >> $output
 
@@ -368,8 +406,44 @@
 
 mv $log $outdir/log.html
 
-cp $outdir/index.html $log
+echo "<html><center><h1><a href='index.html'>Click here for the results</a></h1>Tip: Open it in a new tab (middle mouse button or right mouse button -> 'open in new tab' on the link above)<br />" > $log
+echo "<table border = 1>" >> $log
+echo "<thead><tr><th>Info</th><th>Sequences</th><th>Percentage</th></tr></thead>" >> $log
+tIFS="$TMP"
+IFS=$'\t'
+while read step seq perc
+	do
+		echo "<tr>" >> $log
+		echo "<td>$step</td>" >> $log
+		echo "<td>$seq</td>" >> $log
+		echo "<td>${perc}%</td>" >> $log
+		echo "</tr>" >> $log
+done < $outdir/filtering_steps.txt
+echo "</table border></center></html>" >> $log
+
+IFS="$tIFS"
+
 
 echo "---------------- Done! ----------------"
 echo "---------------- Done! ----------------<br />" >> $outdir/log.html
 
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+