Mercurial > repos > davidvanzessen > argalaxy_tools

comparison report_clonality/r_wrapper.sh @ 58:a073fa12ef98 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author davidvanzessen
date Fri, 18 Mar 2016 08:02:22 -0400
parents
children 55f18bf19d72
comparison
equal deleted inserted replaced
57:16c7fc1c4bf8 58:a073fa12ef98
1 #!/bin/bash
2
3 inputFile=$1
4 outputDir=$3
5 outputFile=$3/index.html #$2
6 clonalType=$4
7 species=$5
8 locus=$6
9 filterproductive=$7
10 clonality_method=$8
11
12 dir="$(cd "$(dirname "$0")" && pwd)"
13 useD="false"
14 if grep -q "$species.*${locus}D" "$dir/genes.txt" ; then
15 echo "species D region in reference db"
16 useD="true"
17 fi
18 echo "$species"
19 if [[ "$species" == *"custom"* ]] ; then
20 loci=(${locus//;/ })
21 useD="true"
22 echo "${loci[@]}"
23 if [[ "${#loci[@]}" -eq "2" ]] ; then
24 useD="false"
25 fi
26 fi
27 mkdir $3
28 cp $dir/genes.txt $outputDir
29 Rscript --verbose $dir/RScript.r $inputFile $outputDir $outputDir $clonalType "$species" "$locus" $filterproductive ${clonality_method} 2>&1
30 cp $dir/tabber.js $outputDir
31 cp $dir/style.css $outputDir
32 cp $dir/script.js $outputDir
33 cp $dir/jquery-1.11.0.min.js $outputDir
34 samples=`cat $outputDir/samples.txt`
35
36 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 />" > $2
37 echo "<table border = 1>" >> $2
38 echo "<thead><tr><th>Sample/Replicate</th><th>All</th><th>Productive</th><th>Unique Productive</th><th>Unproductive</th><th>Unique Unproductive</th></tr></thead>" >> $2
39 while IFS=, read sample all productive perc_prod productive_unique perc_prod_un unproductive perc_unprod unproductive_unique perc_unprod_un
40 do
41 echo "<tr><td>$sample</td>" >> $2
42 echo "<td>$all</td>" >> $2
43 echo "<td>$productive (${perc_prod}%)</td>" >> $2
44 echo "<td>$productive_unique (${perc_prod_un}%)</td>" >> $2
45 echo "<td>$unproductive (${perc_unprod}%)</td>" >> $2
46 echo "<td>$unproductive_unique (${perc_unprod_un}%)</td></tr>" >> $2
47 done < $outputDir/productive_counting.txt
48 echo "</table border></center></html>" >> $2
49
50 echo "productive_counting.txt"
51 echo "<html><head><title>Report on:" >> $outputFile
52
53 mkdir $outputDir/circos
54 cp $dir/circos/* $outputDir/circos/
55 CIRCOSTOOLS="/data/galaxy/galaxy-dist/toolsheddependencies/circos/0.64/saskia-hiltemann/cg_circos_plots/bbfdd52d64fd/circos-tools-0.21/tools"
56 CIRCOSDIR="/data/galaxy/galaxy-dist/toolsheddependencies/circos/0.64/saskia-hiltemann/cg_circos_plots/bbfdd52d64fd/bin/"
57 USECIRCOS="no"
58 if [ -d "$CIRCOSDIR" ]; then
59 USECIRCOS="yes"
60 fi
61 echo "Using Circos: $USECIRCOS"
62 sed -i "s%DATA_DIR%$outputDir/circos%" $outputDir/circos/circos.conf
63 for sample in $samples; do #output the samples to a file and create the circos plots with the R script output
64 echo " $sample" >> $outputFile
65
66 if [[ "$USECIRCOS" != "yes" ]]; then
67 continue
68 fi
69
70 circos_file="$outputDir/${sample}_VJ_circos.txt"
71 echo -e -n "labels$(cat ${circos_file})" > ${circos_file}
72 cat "${circos_file}" | $CIRCOSTOOLS/tableviewer/bin/parse-table -configfile $dir/circos/parse-table.conf 2>&1 | $CIRCOSTOOLS/tableviewer/bin/make-conf -dir $outputDir/circos/
73 $CIRCOSDIR/circos -conf $outputDir/circos/circos.conf 2>&1
74 mv $outputDir/circos/circos.png $outputDir/circosVJ_${sample}.png
75
76
77 if [[ "$useD" == "true" ]] ; then
78 circos_file="$outputDir/${sample}_VD_circos.txt"
79 echo -e -n "labels$(cat ${circos_file})" > ${circos_file}
80 cat "${circos_file}" | $CIRCOSTOOLS/tableviewer/bin/parse-table -configfile $dir/circos/parse-table.conf 2>&1 | $CIRCOSTOOLS/tableviewer/bin/make-conf -dir $outputDir/circos/
81 $CIRCOSDIR/circos -conf $outputDir/circos/circos.conf 2>&1
82 mv $outputDir/circos/circos.png $outputDir/circosVD_${sample}.png
83
84 circos_file="$outputDir/${sample}_DJ_circos.txt"
85 echo -e -n "labels$(cat ${circos_file})" > ${circos_file}
86 cat "${circos_file}" | $CIRCOSTOOLS/tableviewer/bin/parse-table -configfile $dir/circos/parse-table.conf 2>&1 | $CIRCOSTOOLS/tableviewer/bin/make-conf -dir $outputDir/circos/
87 $CIRCOSDIR/circos -conf $outputDir/circos/circos.conf 2>&1
88 mv $outputDir/circos/circos.png $outputDir/circosDJ_${sample}.png
89
90 fi
91 done
92 echo "</title><script type='text/javascript' src='jquery-1.11.0.min.js'></script>" >> $outputFile
93 echo "<script type='text/javascript' src='tabber.js'></script>" >> $outputFile
94 echo "<script type='text/javascript' src='script.js'></script>" >> $outputFile
95 echo "<link rel='stylesheet' type='text/css' href='style.css'></head>" >> $outputFile
96 echo "<div class='tabber'><div class='tabbertab' title='Gene frequencies'>" >> $outputFile
97
98 echo "<img src='CDR3LengthPlot.png'/><br />" >> $outputFile
99 echo "<img src='VFPlot.png'/>" >> $outputFile
100 if [[ "$useD" == "true" ]] ; then
101 echo "<img src='DFPlot.png'/>" >> $outputFile
102 fi
103 echo "<img src='JFPlot.png'/>" >> $outputFile
104 echo "<img src='VPlot.png'/>" >> $outputFile
105 if [[ "$useD" == "true" ]] ; then
106 echo "<img src='DPlot.png'/>" >> $outputFile
107 fi
108 echo "<img src='JPlot.png'/>" >> $outputFile
109 echo "<img src='AAComposition.png'/></div>" >> $outputFile
110
111 #Heatmaps
112
113 count=1
114 echo "<div class='tabbertab' title='Heatmaps'><div class='tabber'>" >> $outputFile
115 for sample in $samples; do
116 echo "<div class='tabbertab' title='$sample'><table border='1'><tr>" >> $outputFile
117 if [[ "$useD" == "true" ]] ; then
118 echo "<td><img src='HeatmapVD_$sample.png'/></td>" >> $outputFile
119 fi
120 echo "<td><img src='HeatmapVJ_$sample.png'/></td>" >> $outputFile
121 if [[ "$useD" == "true" ]] ; then
122 echo "<td><img src='HeatmapDJ_$sample.png'/></td>" >> $outputFile
123 fi
124 echo "</tr></table></div>" >> $outputFile
125 count=$((count+1))
126 done
127 echo "</div></div>" >> $outputFile
128
129 #circos
130
131 if [[ "$USECIRCOS" == "yes" ]]; then
132
133 echo "<div class='tabbertab' title='Circos'><div class='tabber'>" >> $outputFile
134 for sample in $samples; do
135 echo "<div class='tabbertab' title='$sample'><table border='1'><center>" >> $outputFile
136 if [[ "$useD" == "true" ]] ; then
137 echo "<tr><td>V-D</td><td><img src='circosVD_${sample}.png' width='700' height='700'/></td></tr>" >> $outputFile
138 fi
139 echo "<tr><td>V-J</td><td><img src='circosVJ_${sample}.png' width='700' height='700'/></td></tr>" >> $outputFile
140 if [[ "$useD" == "true" ]] ; then
141 echo "<tr><td>D-J</td><td><img src='circosDJ_${sample}.png' width='700' height='700'/></td></tr>" >> $outputFile
142 fi
143 echo "<center></table></div>" >> $outputFile
144 count=$((count+1))
145 done
146 echo "</div></div>" >> $outputFile
147 fi
148 #echo "<div class='tabbertab' title='Interactive'><svg class='chart'></svg><script src='http://d3js.org/d3.v3.min.js'></script></div>" >> $outputFile
149
150 hasReplicateColumn="$(if head -n 1 $inputFile | grep -q 'Replicate'; then echo 'Yes'; else echo 'No'; fi)"
151 echo "$hasReplicateColumn"
152 #if its a 'new' merged file with replicate info
153 if [[ "$hasReplicateColumn" == "Yes" ]] ; then
154 echo "<div class='tabbertab' title='Clonality'><div class='tabber'>" >> $outputFile
155 for sample in $samples; do
156 echo "${clonality_method}"
157 if [[ "${clonality_method}" == "old" ]] ; then
158 echo "in old"
159 clonalityScore="$(cat $outputDir/ClonalityScore_$sample.csv)"
160 echo "<div class='tabbertab' title='$sample'><table border='1'>" >> $outputFile
161 echo "<tr><td colspan='4'>Clonality Score: $clonalityScore</td></tr>" >> $outputFile
162
163 #replicate,reads,squared
164 echo "<tr><td>Replicate ID</td><td>Number of Reads</td><td>Reads Squared</td><td></td></tr>" >> $outputFile
165 while IFS=, read replicate reads squared
166 do
167
168 echo "<tr><td>$replicate</td><td>$reads</td><td>$squared</td><td></td></tr>" >> $outputFile
169 done < $outputDir/ReplicateReads_$sample.csv
170
171 #sum of reads and reads squared
172 while IFS=, read readsSum squaredSum
173 do
174 echo "<tr><td>Sum</td><td>$readsSum</td><td>$squaredSum</td></tr>" >> $outputFile
175 done < $outputDir/ReplicateSumReads_$sample.csv
176
177 #overview
178 echo "<tr><td>Coincidence Type</td><td>Raw Coincidence Freq</td><td>Coincidence Weight</td><td>Coincidences, Weighted</td></tr>" >> $outputFile
179 while IFS=, read type count weight weightedCount
180 do
181 echo "<tr><td>$type</td><td>$count</td><td>$weight</td><td>$weightedCount</td></tr>" >> $outputFile
182 done < $outputDir/ClonalityOverView_$sample.csv
183 echo "</table></div>" >> $outputFile
184 else
185 echo "in new"
186 clonalityScore="$(cat $outputDir/lymphclon_clonality_${sample}.csv)"
187 echo "<div class='tabbertab' title='$sample'>" >> $outputFile
188 echo "Lymphclon clonality score: <br />$clonalityScore<br /><br />" >> $outputFile
189 echo "<table border = 1>" >> $outputFile
190 while IFS=, read type count
191 do
192 echo "<tr><td>$type</td><td>$count</td></tr>" >> $outputFile
193 done < $outputDir/lymphclon_coincidences_$sample.csv
194 echo "</table></div>" >> $outputFile
195 fi
196 done
197 echo "</div></div>" >> $outputFile
198 fi
199
200 #hasJunctionData="$(if head -n 1 $inputFile | grep -qE '3V.REGION.trimmed.nt.nb'; then echo 'Yes'; else echo 'No'; fi)"
201
202 #if [[ "$hasJunctionData" == "Yes" ]] ; then
203 if [ -a "$outputDir/junctionAnalysisProd_mean.csv" ] ; then
204 echo "<div class='tabbertab' title='Junction Analysis'>" >> $outputFile
205 echo "<table border='1' id='junction_table'> <caption>Productive mean</caption><thead><tr><th>Sample</th><th>count</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>" >> $outputFile
206 while IFS=, read Sample unique VHDEL P1 N1 P2 DELDH DHDEL P3 N2 P4 DELJH TotalDel TotalN TotalP
207 do
208 echo "<tr><td>$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
209 done < $outputDir/junctionAnalysisProd_mean.csv
210 echo "</tbody></table>" >> $outputFile
211
212 echo "<table border='1' id='junction_table'> <caption>Unproductive mean</caption><thead><tr><th>Sample</th><th>count</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>" >> $outputFile
213 while IFS=, read Sample unique VHDEL P1 N1 P2 DELDH DHDEL P3 N2 P4 DELJH TotalDel TotalN TotalP
214 do
215 echo "<tr><td>$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
216 done < $outputDir/junctionAnalysisUnProd_mean.csv
217 echo "</tbody></table>" >> $outputFile
218
219 echo "<table border='1' id='junction_table'> <caption>Productive median</caption><thead><tr><th>Sample</th><th>count</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>" >> $outputFile
220 while IFS=, read Sample unique VHDEL P1 N1 P2 DELDH DHDEL P3 N2 P4 DELJH TotalDel TotalN TotalP
221 do
222 echo "<tr><td>$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
223 done < $outputDir/junctionAnalysisProd_median.csv
224 echo "</tbody></table>" >> $outputFile
225
226 echo "<table border='1' id='junction_table'> <caption>Unproductive median</caption><thead><tr><th>Sample</th><th>count</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>" >> $outputFile
227 while IFS=, read Sample unique VHDEL P1 N1 P2 DELDH DHDEL P3 N2 P4 DELJH TotalDel TotalN TotalP
228 do
229 echo "<tr><td>$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
230 done < $outputDir/junctionAnalysisUnProd_median.csv
231 echo "</tbody></table>" >> $outputFile
232
233 echo "</div>" >> $outputFile
234 fi
235
236 echo "<div class='tabbertab' title='Comparison'><table border='1'><tr><th>ID</th><th>Include</th></tr>" >> $outputFile
237 for sample in $samples; do
238 echo "<tr><td>$sample</td><td><input type='checkbox' onchange=\"javascript:compareAdd('$sample')\" id='compare_checkbox_$sample'/></td></tr>" >> $outputFile
239 done
240 echo "</table><div name='comparisonarea'>" >> $outputFile
241 echo "<table><tr id='comparison_table_vd'></tr></table>" >> $outputFile
242 echo "<table><tr id='comparison_table_vj'></tr></table>" >> $outputFile
243 echo "<table><tr id='comparison_table_dj'></tr></table>" >> $outputFile
244 echo "</div></div>" >> $outputFile
245
246 echo "<div class='tabbertab' title='Downloads'>" >> $outputFile
247 echo "<table border='1'>" >> $outputFile
248 echo "<tr><th>Description</th><th>Link</th></tr>" >> $outputFile
249 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
250 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
251
252 echo "<tr><td>The dataset used to generate the CDR3 length frequency graph</td><td><a href='CDR3LengthPlot.csv'>Download</a></td></tr>" >> $outputFile
253
254 echo "<tr><td>The dataset used to generate the V gene family frequency graph</td><td><a href='VFFrequency.csv'>Download</a></td></tr>" >> $outputFile
255 if [[ "$useD" == "true" ]] ; then
256 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
257 fi
258 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
259
260 echo "<tr><td>The dataset used to generate the V gene frequency graph</td><td><a href='VFrequency.csv'>Download</a></td></tr>" >> $outputFile
261 if [[ "$useD" == "true" ]] ; then
262 echo "<tr><td>The dataset used to generate the D gene frequency graph</td><td><a href='DFrequency.csv'>Download</a></td></tr>" >> $outputFile
263 fi
264 echo "<tr><td>The dataset used to generate the J gene frequency graph</td><td><a href='JFrequency.csv'>Download</a></td></tr>" >> $outputFile
265 echo "<tr><td>The dataset used to generate the AA composition graph</td><td><a href='AAComposition.csv'>Download</a></td></tr>" >> $outputFile
266
267 for sample in $samples; do
268 if [[ "$useD" == "true" ]] ; then
269 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
270 fi
271 echo "<tr><td>The data used to generate the VJ heatmap for $sample.</td><td><a href='HeatmapVJ_$sample.csv'>Download</a></td></tr>" >> $outputFile
272 if [[ "$useD" == "true" ]] ; then
273 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
274 fi
275 done
276
277 echo "</table>" >> $outputFile
278 echo "</div></html>" >> $outputFile