Mercurial > repos > davidvanzessen > combined_immune_repertoire_pipeline

diff combined.sh @ 9:8d83319a0f3d draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author davidvanzessen
date Tue, 10 Dec 2013 05:53:08 -0500
parents e6f28d0a0f14
children
line wrap: on
line diff
--- a/combined.sh	Mon Nov 25 09:21:55 2013 -0500
+++ b/combined.sh	Tue Dec 10 05:53:08 2013 -0500
@@ -1,55 +1,42 @@
 #!/bin/bash
 
-#export IGDATA=/home/david/Galaxy/galaxy-dist/toolsheddependencies/igBlastn/1.0.0/davidvanzessen/igblast_human/1c64c977624e/ncbi-igblast-1.0.0/;
+export IGDATA=/home/david/tmp/ncbi-igblast-1.0.0;
 
 clonalType=${@:(-3):1}
 html=${@:(-2):1}
 imageDir=${@:(-1):1}
+dataCount=`expr $# - 3`
+inputData=${@:(1):dataCount}
 dir="$(cd "$(dirname "$0")" && pwd)"
-fileCount=`expr $# - 3`
 array=("$@")
 echo "<html><h3>Progress</h3><table><tr><td>info</td></tr>" > $html
 echo "<tr><td>-----------------------------------</td></tr>" >> $html
-limit=`expr $fileCount / 2`
 
-function blastAndParse {
-	echo "<tr><td>Starting blast of $2</td></tr>" >> $html
-	fileName=$(basename $1)
-	$IGDATA/bin/igblastn -germline_db_V $IGDATA/database/human_gl_V -germline_db_J $IGDATA/database/human_gl_J -germline_db_D $IGDATA/database/human_gl_D -domain_system imgt -query $1 -auxiliary_data $IGDATA/optional_file/human_gl.aux -show_translation -outfmt 3 > $PWD/$fileName
-	echo "<tr><td>Finished blast of $2</td></tr>" >> $html
-	
-	echo "<tr><td>Starting parse of $2</td></tr>" >> $html
-	parsedFileName="${fileName%.*}"
-	parsedFileName="$parsedFileName.parsed"
-	perl $dir/igparse.pl $PWD/$fileName 0 | grep -v "D:" | cut -f2- > $parsedFileName
-	echo "<tr><td>Finished parse of $2</td></tr>" >> $html
-}
-
-for ((i=0;i<$fileCount;i=$((i+2))))
-do
-	next=$((i+1))
-	blastAndParse ${array[$i]} ${array[$next]} &
+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 $current)
+    convertedFileName="${fileName%.*}"
+    convertedFileName="$PWD/$convertedFileName.converted"
+	bash $dir/imgtconvert.sh $current $id $count $convertedFileName
+	mergerInput+=($convertedFileName)
+	count=$((count+1))
 done
-wait
 
 
 
 echo "<tr><td>-----------------------------------</td></tr>" >> $html
 echo "<tr><td>merging</td></tr>" >> $html
 
-count=0
-for ((i=0;i<$fileCount;i=$((i+2))))
-do
-	id=$((i+1))
-	place=$((count+limit))
-	fn=$(basename ${array[$i]})
-	fn="${fn%.*}"
-	mergeInputs[$count]="$PWD/$fn.parsed"
-	mergeIDs[$place]=${array[$id]}
-	count=$((count+1))
-done
-
-python $dir/igblastmerge.py --input ${mergeInputs[*]} --id ${mergeIDs[*]} --output $PWD/merged.txt
+python $dir/igblastmerge.py ${mergerInput[*]}  --output $PWD/merged.txt
 
 echo "<tr><td>done</td></tr>" >> $html
 echo "<tr><td>-----------------------------------</td></tr>" >> $html
@@ -68,10 +55,43 @@
 
 samples=`cat $outputDir/samples.txt`
 count=1
-echo "<table border='1'><caption><h3>$clonalType</h3></caption>" >> $outputFile
+echo "<table border='1'><caption><a href='allUnique.tsv'><h3>$clonalType</h3></a></caption>" >> $outputFile
+hasReplicateColumn="$(if head -n 1 $inputFile | grep -q 'Replicate'; then echo 'Yes'; else echo 'No'; fi)"
 for sample in $samples; do
-	echo "<tr><td colspan='3' height='100'></td>" >> $outputFile
+	clonalityScore="$(cat $outputDir/ClonalityScore_$sample.csv)"
+	echo "<tr><td colspan='3' height='100'></td></tr>" >> $outputFile
 	echo "<tr><td colspan='3'><h1>$sample</h1></td></tr>" >> $outputFile
+
+	echo "$hasReplicateColumn"
+	#if its a 'new' merged file with replicate info
+	if [[ "$hasReplicateColumn" == "Yes" ]] ; then
+		echo "<tr><td colspan='3'><a href='clonality_$sample.tsv'><h2>Clonality Score: $clonalityScore</h2></a></td></tr>" >> $outputFile
+
+		#replicate,reads,squared
+		echo "<tr><td colspan='3'><table border='1'><tr><th>Replicate ID</th><th>Number of Reads</th><th>Reads Squared</th></tr>" >> $outputFile
+		while IFS=, read replicate reads squared
+		do
+
+			echo "<tr><td><a href='clonality_${sample}_$replicate.tsv'>$replicate</a></td><td>$reads</td><td>$squared</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
+
+		echo "</table></td></tr>" >> $outputFile
+
+		#overview
+		echo "<tr><td colspan='3'><table border='1'><tr><th>Coincidence Type</th><th>Raw Coincidence Freq</th><th>Coincidence Weight</th><th>Coincidences, Weighted</th></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></td></tr>" >> $outputFile
+	fi
+
 	echo "<tr><td><h2>V-D Heatmap:</h2></td><td><h2>V-J Heatmap:</h2></td><td><h2>D-J Heatmap:</h2></td></tr><tr>" >> $outputFile
 	mv "$outputDir/HeatmapVD_$sample.png" "$outputDir/VD_$sample.png"
 	echo "<td><img src='VD_$sample.png'/></td>" >> $outputFile
@@ -83,4 +103,4 @@
 done
 echo "</table>" >> $outputFile
 
-echo "</html>" >> $2
+echo "</html>" >> $outputFile