Mercurial > repos > davidvanzessen > complete_immunerepertoire_igg

diff complete.sh @ 0:7d97fa9a0423 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author davidvanzessen
date Fri, 09 May 2014 09:35:32 -0400
parents
children 778a9d130904
line wrap: on
line diff
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/complete.sh	Fri May 09 09:35:32 2014 -0400
@@ -0,0 +1,204 @@
+#!/bin/bash
+
+clonalType=${@:(-6):1}
+html=${@:(-5):1}
+imageDir=${@:(-4):1}
+species=${@:(-3):1}
+locus=${@:(-2):1}
+selection=${@:(-1):1}
+dataCount=`expr $# - 6`
+inputData=${@:(1):dataCount}
+echo ${inputData[@]}
+dir="$(cd "$(dirname "$0")" && pwd)"
+array=("$@")
+echo "<html><h3>Progress</h3><table><tr><td>info</td></tr>" > $html
+echo "<tr><td>-----------------------------------</td></tr>" >> $html
+
+mkdir $PWD/igblastdatabase
+unzip $dir/database.zip -d $PWD/igblastdatabase/
+export IGDATA=$PWD/igblastdatabase/
+
+function blastAndParse {
+	echo "<tr><td>Starting blast of sample $3 of patient $2</td></tr>" >> $html
+	echo "igblastn -germline_db_V $PWD/igblastdatabase/database/human_gl_V -germline_db_J $PWD/igblastdatabase/database/human_gl_J -germline_db_D $PWD/igblastdatabase/database/human_gl_D -domain_system imgt -query $1 -auxiliary_data $PWD/igblastdatabase/optional_file/human_gl.aux -show_translation -outfmt 3 > $PWD/$4"
+	/home/galaxy/galaxy/igblast/igblastn -germline_db_V $PWD/igblastdatabase/database/human_gl_V -germline_db_J $PWD/igblastdatabase/database/human_gl_J -germline_db_D $PWD/igblastdatabase/database/human_gl_D -domain_system imgt -query $1 -auxiliary_data $PWD/igblastdatabase/optional_file/human_gl.aux -show_translation -outfmt 3 > $PWD/$4
+	echo "<tr><td>Finished blast of sample $3 of patient $2</td></tr>" >> $html
+
+	echo "<tr><td>Starting parse of sample $3 of patient $2</td></tr>" >> $html
+	perl $dir/igparse.pl $PWD/$4 0 | grep -v "D:" | cut -f2- > $5
+	echo "<tr><td>Finished parse of sample $3 of patient $2</td></tr>" >> $html
+}
+
+function imgtConvert {
+	echo "<tr><td>Starting imgt convert of sample $3 of patient $2</td></tr>" >> $html
+	bash $dir/imgtconvert.sh $1 $2 $3 $4
+	echo "<tr><td>Finished conversion of sample $3 of patient $2</td></tr>" >> $html
+	
+}
+
+id=${inputData[0]}
+forwardSlash="/"
+mergerInput=()
+count=0
+for current in $inputData; do
+    if [[ "$current" != *"$forwardSlash"* ]]; then
+        id=$current
+        count=0
+        mergerInput+=($id)
+        continue
+    fi
+	fileName=$(basename $current)
+	parsedFileName="${fileName%.*}"
+	parsedFileName="$PWD/$parsedFileName.parsed"
+	f=$(file $current)
+	zipType="Zip archive"
+  if [[ "$f" == *"$zipType"* ]]
+	then
+		echo "<tr><td>Sample $count of patient $id is a zip file, using IMGT Loader</td></tr>" >> $html
+	  fileName=$(basename $current)
+		imgtConvert $current $id $count $parsedFileName &
+	else
+		echo "<tr><td>Sample $count of patient $id is not a zip file, using igBLASTn</td></tr>" >> $html
+		blastAndParse $current $id $count $fileName $parsedFileName &
+	fi
+	mergerInput+=($parsedFileName)
+	count=$((count+1))
+done
+wait
+
+
+
+echo "<tr><td>-----------------------------------</td></tr>" >> $html
+echo "<tr><td>merging</td></tr>" >> $html
+
+python $dir/igblastmerge.py ${mergerInput[*]}  --output $PWD/merged.txt
+
+echo "<tr><td>done</td></tr>" >> $html
+echo "<tr><td>-----------------------------------</td></tr>" >> $html
+echo "<tr><td>plotting</td></tr>" >> $html
+
+
+inputFile=$PWD/merged.txt
+outputDir=$imageDir
+outputFile=$outputDir/index.html
+mkdir $outputDir
+Rscript --verbose $dir/RScript.r $inputFile $outputDir $outputDir $clonalType $species $locus $selection 2>&1
+cp $dir/tabber.js $outputDir
+cp $dir/style.css $outputDir
+cp $dir/script.js $outputDir
+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)</center></html>" > $html
+echo "<html><head><title>Report on:" >> $outputFile
+for sample in $samples; do
+	echo " $sample" >> $outputFile
+done
+echo "</title><script type='text/javascript' src='jquery-1.11.0.min.js'></script>" >> $outputFile
+echo "<script type='text/javascript' src='tabber.js'></script>" >> $outputFile
+echo "<script type='text/javascript' src='script.js'></script>" >> $outputFile
+echo "<script type='text/javascript' src='jquery.tablesorter.min.js'></script>" >> $outputFile
+echo "<link rel='stylesheet' type='text/css' href='style.css'></head>" >> $outputFile
+
+echo "<div class='tabber'><div class='tabbertab' title='Gene frequencies'>" >> $outputFile
+echo "<img src='CDR3LengthPlot.png'/><br />" >> $outputFile
+echo "<img src='VFPlot.png'/>" >> $outputFile
+echo "<img src='DFPlot.png'/>" >> $outputFile
+echo "<img src='JFPlot.png'/>" >> $outputFile
+echo "<img src='VPlot.png'/>" >> $outputFile
+echo "<img src='DPlot.png'/>" >> $outputFile
+echo "<img src='JPlot.png'/></div>" >> $outputFile
+
+samples=`cat $outputDir/samples.txt`
+count=1
+echo "<div class='tabbertab' title='Heatmaps'><div class='tabber'>" >> $outputFile
+for sample in $samples; do
+	echo "<div class='tabbertab' title='$sample'><table border='1'><tr>" >> $outputFile
+	
+	mv "$outputDir/HeatmapVD_$sample.png" "$outputDir/VD_$sample.png"
+	echo "<td><img src='VD_$sample.png'/></td>" >> $outputFile
+	mv "$outputDir/HeatmapVJ_$sample.png" "$outputDir/VJ_$sample.png"
+	echo "<td><img src='VJ_$sample.png'/></td>" >> $outputFile
+	mv "$outputDir/HeatmapDJ_$sample.png" "$outputDir/DJ_$sample.png"
+	echo "<td><img src='DJ_$sample.png'/></td></tr></table></div>" >> $outputFile
+	count=$((count+1))
+done
+echo "</div></div>" >> $outputFile
+
+
+hasReplicateColumn="$(if head -n 1 $inputFile | grep -q 'Replicate'; then echo 'Yes'; else echo 'No'; fi)"
+#if its a 'new' merged file with replicate info
+if [[ "$hasReplicateColumn" == "Yes" ]] ; then
+	echo "<div class='tabbertab' title='Clonality'><div class='tabber'>" >> $outputFile
+	for sample in $samples; do
+		clonalityScore="$(cat $outputDir/ClonalityScore_$sample.csv)"
+		echo "<div class='tabbertab' title='$sample'><table border='1'>" >> $outputFile
+		echo "<tr><td colspan='4'>Clonality Score: $clonalityScore</td></tr>" >> $outputFile
+
+		#replicate,reads,squared
+		echo "<tr><td>Replicate ID</td><td>Number of Reads</td><td>Reads Squared</td><td></td></tr>" >> $outputFile
+		while IFS=, read replicate reads squared
+		do
+			
+			echo "<tr><td>$replicate</td><td>$reads</td><td>$squared</td><td></td></tr>" >> $outputFile
+		done < $outputDir/ReplicateReads_$sample.csv
+		
+		#sum of reads and reads squared
+		while IFS=, read readsSum squaredSum
+			do
+				echo "<tr><td>Sum</td><td>$readsSum</td><td>$squaredSum</td></tr>" >> $outputFile
+		done < $outputDir/ReplicateSumReads_$sample.csv
+		
+		#overview
+		echo "<tr><td>Coincidence Type</td><td>Raw Coincidence Freq</td><td>Coincidence Weight</td><td>Coincidences, Weighted</td></tr>" >> $outputFile
+		while IFS=, read type count weight weightedCount
+		do
+			echo "<tr><td>$type</td><td>$count</td><td>$weight</td><td>$weightedCount</td></tr>" >> $outputFile
+		done < $outputDir/ClonalityOverView_$sample.csv
+		echo "</table></div>" >> $outputFile
+	done
+	echo "</div></div>" >> $outputFile
+fi
+
+hasJunctionData="$(if head -n 1 $inputFile | grep -q '3V-REGION trimmed-nt nb'; then echo 'Yes'; else echo 'No'; fi)"
+
+if [[ "$hasJunctionData" == "Yes" ]] ; then
+	echo "<div class='tabbertab' title='Junction Analysis'><table border='1' id='junction_table'  class='tablesorter'><thead><tr><th>Sample</th><th>unique</th><th>VH.DEL</th><th>P1</th><th>N1</th><th>P2</th><th>DEL.DH</th><th>DH.DEL</th><th>P3</th><th>N2</th><th>P4</th><th>DEL.JH</th><th>Total.Del</th><th>Total.N</th><th>Total.P</th><thead></tr><tbody>" >> $outputFile
+	while IFS=, read Sample unique VHDEL P1 N1 P2 DELDH DHDEL P3 N2 P4 DELJH TotalDel TotalN TotalP
+	do
+		echo "<tr><td>$Sample</td><td>$unique</td><td>$VHDEL</td><td>$P1</td><td>$N1</td><td>$P2</td><td>$DELDH</td><td>$DHDEL</td><td>$P3</td><td>$N2</td><td>$P4</td><td>$DELJH</td><td>$TotalDel</td><td>$TotalN</td><td>$TotalP</td></tr>" >> $outputFile
+	done < $outputDir/junctionAnalysis.csv
+	echo "</tbody></table></div>" >> $outputFile
+fi
+
+echo "<div class='tabbertab' title='Comparison'><table border='1'><tr><th>ID</th><th>Include</th></tr>" >> $outputFile
+for sample in $samples; do
+	echo "<tr><td>$sample</td><td><input type='checkbox' onchange=\"javascript:compareAdd('$sample')\" id='compare_checkbox_$sample'/></td></tr>" >> $outputFile
+done
+echo "</table><div name='comparisonarea'>" >> $outputFile
+echo "<table><tr id='comparison_table_vd'></tr></table>" >> $outputFile
+echo "<table><tr id='comparison_table_vj'></tr></table>" >> $outputFile
+echo "<table><tr id='comparison_table_dj'></tr></table>" >> $outputFile
+echo "</div></div>" >> $outputFile
+
+echo "<div class='tabbertab' title='Downloads'>" >> $outputFile
+echo "<table border='1'>" >> $outputFile
+echo "<tr><th>Description</th><th>Link</th></tr>" >> $outputFile
+echo "<tr><td>The dataset used to generate the frequency graphs and the heatmaps (Unique based on clonaltype, $clonalType)</td><td><a href='allUnique.csv'>Download</a></td></tr>" >> $outputFile
+echo "<tr><td>The dataset used to calculate clonality score (Unique based on clonaltype, $clonalType)</td><td><a href='clonalityComplete.csv'>Download</a></td></tr>" >> $outputFile
+
+echo "<tr><td>The dataset used to generate the CDR3 length frequency graph</td><td><a href='CDR3LengthPlot.csv'>Download</a></td></tr>" >> $outputFile
+
+echo "<tr><td>The dataset used to generate the V gene family frequency graph</td><td><a href='VFFrequency.csv'>Download</a></td></tr>" >> $outputFile
+echo "<tr><td>The dataset used to generate the D gene family frequency graph</td><td><a href='DFFrequency.csv'>Download</a></td></tr>" >> $outputFile
+echo "<tr><td>The dataset used to generate the J gene family frequency graph</td><td><a href='JFFrequency.csv'>Download</a></td></tr>" >> $outputFile
+
+echo "<tr><td>The dataset used to generate the V gene frequency graph</td><td><a href='VFrequency.csv'>Download</a></td></tr>" >> $outputFile
+echo "<tr><td>The dataset used to generate the D gene frequency graph</td><td><a href='DFrequency.csv'>Download</a></td></tr>" >> $outputFile
+echo "<tr><td>The dataset used to generate the J gene frequency graph</td><td><a href='JFrequency.csv'>Download</a></td></tr>" >> $outputFile
+
+for sample in $samples; do
+	echo "<tr><td>The data used to generate the VD heatmap for $sample.</td><td><a href='HeatmapVD_$sample.csv'>Download</a></td></tr>" >> $outputFile
+	echo "<tr><td>The data used to generate the VJ heatmap for $sample.</td><td><a href='HeatmapVJ_$sample.csv'>Download</a></td></tr>" >> $outputFile
+	echo "<tr><td>The data used to generate the DJ heatmap for $sample.</td><td><a href='HeatmapDJ_$sample.csv'>Download</a></td></tr>" >> $outputFile
+done
+
+echo "</table>" >> $outputFile
+echo "</div></html>" >> $outputFile