18
|
1 #!usr/bin/bash
|
|
2
|
|
3 #########################################################################################################
|
|
4 # ARGUMENTS FROM alfa_wrapper.xml #
|
|
5 #########################################################################################################
|
19
|
6 galaxyRoot=$1;
|
|
7 toolDir=$2
|
|
8 configFile=$3;
|
|
9 logReport=$4;
|
18
|
10 sed -i -e '/^$/d; s/\t//g;' $configFile;
|
|
11 printf "__________________________________________________________________\n\n" > $logReport
|
|
12 printf " ALFA CONFIG \n" >> $logReport
|
|
13 printf "__________________________________________________________________\n" >> $logReport
|
|
14 cat $configFile >> $logReport
|
|
15
|
|
16 #########################################################################################################
|
|
17 # INITIALIZATION OF THE VARIABLES from $configFile #
|
|
18 #########################################################################################################
|
|
19 #_INPUT1
|
|
20 annotationSource=`grep -P '^annotationSource ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
21 if [ "$annotationSource" == "personal_gtf" ]; then
|
|
22 annotationFile=`grep -P '^annotationFile ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
23 elif [ "$annotationSource" == "built_in_index" ]; then
|
|
24 built_in_index_prefix=`grep -P '^built_in_index_prefix ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
25 else
|
|
26 strandedIndex=`grep -P '^strandedIndex ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
27 unstrandedIndex=`grep -P '^unstrandedIndex ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
28 fi
|
|
29
|
|
30 #_INPUT2
|
|
31 readsType=`grep -P '^readsType ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
32 readsFileList=`grep -P '^readsFile\[[0-9]+\] ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
33 readsLabelList=`grep -P '^readsLabel\[[0-9]+\] ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
34
|
|
35 #_OUTPUT CHOICES
|
|
36 plotChoice=`grep -P '^plotChoice ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
37 countFileChoice=`grep -P '^countFileChoice ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
38 indexChoice=`grep -P '^indexChoice ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
39
|
|
40 #_OUTPUT OPTIONS
|
|
41 strandness=`grep -P '^strandness ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
42 categoriesDepth=`grep -P '^categoriesDepth ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
43 plotFormat=`grep -P '^plotFormat ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
44 plotThresholdChoice=`grep -P '^plotThresholdChoice ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
45 if [ "$plotThresholdChoice" == "True" ]; then
|
|
46 yMin=`grep -P '^yMin ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
47 yMax=`grep -P '^yMax ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
48 fi
|
|
49
|
|
50 #_OUTPUT FILES
|
|
51 if [ "$plotChoice" == "True" ]; then
|
|
52 if [ "$plotFormat" == "pdf" ]; then
|
|
53 outputPdf=`grep -P '^outputPdf ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
54 elif [ "$plotFormat" == "svg" ]; then
|
|
55 outputCategoriesSvg=`grep -P '^outputCategoriesSvg ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
56 outputBiotypesSvg=`grep -P '^outputBiotypesSvg ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
57 else
|
|
58 outputCategoriesPng=`grep -P '^outputCategoriesPng ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
59 outputBiotypesPng=`grep -P '^outputBiotypesPng ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
60 fi
|
|
61 fi
|
|
62 if [ "$countFileChoice" == "True" ]; then
|
|
63 outputCountFile=`grep -P '^outputCountFile ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
64 fi
|
|
65 if [ "$indexChoice" == "True" ]; then
|
|
66 outputStrandedIndex=`grep -P '^outputStrandedIndex ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
67 outputUnstrandedIndex=`grep -P '^outputUnstrandedIndex ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`;
|
|
68 fi
|
|
69
|
|
70 #########################################################################################################
|
|
71 # CREATION OF A TMP DIRECTORY FOR THE OUTPUT FILES OF ALFA AND cd #
|
|
72 #########################################################################################################
|
19
|
73 outputDirectory=`mktemp -d "$galaxyRoot"/database/tmp/tmpXXXXXX`;
|
18
|
74 if [ -d $outputDirectory ]; then
|
|
75 chmod -R ugo+wrx $outputDirectory;
|
|
76 rm -R $outputDirectory;
|
|
77 fi
|
|
78 mkdir $outputDirectory;
|
|
79 chmod -R ugo+wrx $outputDirectory;
|
|
80 cd $outputDirectory;
|
|
81
|
|
82 #########################################################################################################
|
|
83 # TEST OF INPUT1 #
|
|
84 #########################################################################################################
|
|
85 if [ "$annotationSource" == "index" ]; then
|
|
86 #need to copy the files.dat to .*index because ALFA requires the extension ".(un)stranded.index"
|
|
87 index="index"
|
|
88 cp $strandedIndex $index".stranded.index"
|
|
89 cp $unstrandedIndex $index".unstranded.index"
|
|
90 fi
|
|
91
|
|
92 #########################################################################################################
|
|
93 # TEST OF INPUT2 AND DETERMINATION OF PYTHON READS INPUT ARGUMENT #
|
|
94 #########################################################################################################
|
|
95 readsListLen=`echo "$readsFileList" | wc -l`;
|
|
96 readsInput="";
|
|
97 for (( i = 1; i <= readsListLen; i++ )) do
|
|
98 readsFile[$i]=`echo "$readsFileList" | awk -v i=$i 'NR==i'`;
|
|
99 readsLabel[$i]=`echo "$readsLabelList" | awk -v i=$i 'NR==i' | sed -e 's/ /_/g'`;
|
|
100 if [ "$readsType" == "bam" ]; then
|
|
101 bamSorted=`samtools view -H "${readsFile[$i]}" | grep -c 'SO:unsorted'`
|
|
102 if [ "$bamSorted" != "0" ] ; then
|
|
103 samtools sort ${readsFile[$i]} ${readsFile[$i]}
|
|
104 fi
|
|
105 else
|
|
106 #need to copy the file.dat to tmp.bedgraph because ALFA requires the extension ".bedgraph"
|
|
107 bedgraphFile="tmpBedgraph_"$i
|
|
108 cp ${readsFile[$i]} $bedgraphFile".bedgraph"
|
|
109 readsFile[$i]=$bedgraphFile
|
|
110 fi
|
|
111 if [ "${readsLabel[$i]}" == "" ]; then
|
|
112 readsLabel[$i]="sample_""$i";
|
|
113 fi
|
|
114 readsInput=$readsInput" "${readsFile[$i]}" "${readsLabel[$i]};
|
|
115 done
|
|
116
|
|
117 #########################################################################################################
|
|
118 # DETERMINATION OF THE APPROPRIATE SCRIPTS ARGUMENTS #
|
|
119 #########################################################################################################
|
19
|
120 scriptPath="$toolDir/";
|
18
|
121 if [ "$annotationSource" == "index" ]; then
|
|
122 scriptInput="-g $index -i ""$readsInput";
|
|
123 elif [ "$annotationSource" == "built_in_index" ]; then
|
|
124 scriptInput="-g $built_in_index_prefix -i ""$readsInput";
|
|
125 else
|
|
126 scriptInput="-a $annotationFile -i ""$readsInput";
|
|
127 fi
|
|
128 if [ "$readsType" = "bedgraph" ]; then
|
|
129 scriptInput=$scriptInput" --bedgraph";
|
|
130 fi
|
|
131 scriptStrandness="-s "$strandness
|
|
132 scriptCategoriesDepth="-d "$categoriesDepth
|
|
133 if [ "$plotChoice" == "True" ]; then
|
|
134 if [ "$plotFormat" == "pdf" ]; then
|
|
135 scriptPlotOutput="--pdf plotFile.pdf";
|
|
136 else
|
|
137 scriptPlotOutput="--"$plotFormat" plotFile";
|
|
138 fi
|
|
139 if [ "$plotThresholdChoice" == "True" ]; then
|
|
140 scriptPlotOutput=$scriptPlotOutput" -t ""$yMin"" ""$yMax"
|
|
141 fi
|
|
142 else
|
|
143 scriptPlotOutput="--n";
|
|
144 fi
|
|
145
|
|
146 #########################################################################################################
|
|
147 # DISPLAY ALFA PROCESS #
|
|
148 #########################################################################################################
|
|
149 printf "__________________________________________________________________\n\n" >> $logReport
|
|
150 printf " ALFA PROCESS \n" >> $logReport
|
|
151 printf "__________________________________________________________________\n" >> $logReport
|
|
152
|
|
153 if [ "$plotChoice" == "False" ] && [ "$countFileChoice" == "False" ] && [ "$indexChoice" == "False" ]; then
|
|
154 cat <<error 1>&2
|
|
155
|
|
156 No output to return.
|
|
157 Process Aborted
|
|
158 error
|
|
159 exit 0
|
|
160 fi
|
|
161
|
|
162 printf "Command:\n" >> $logReport
|
|
163 echo "python ""$scriptPath"ALFA.py $scriptInput $scriptStrandness $scriptCategoriesDepth $scriptPlotOutput >> $logReport;
|
|
164 printf "\n******************************************************************\n" >> $logReport
|
|
165 printf "Temporary Output Directory:\n" >> $logReport
|
|
166 echo $outputDirectory >> $logReport
|
|
167 printf "\n******************************************************************\n" >> $logReport
|
|
168 printf "ALFA prompt:\n" >> $logReport
|
|
169 python "$scriptPath"ALFA.py $scriptInput $scriptStrandness $scriptCategoriesDepth $scriptPlotOutput >> $logReport 2>errorFile;
|
|
170 printf "\n******************************************************************\n" >> $logReport
|
|
171
|
|
172 #########################################################################################################
|
|
173 # REDIRECTION OF ERRORS - TMP SOURCE ALFA.PY MUST BE CORRECTED SOON #
|
|
174 #########################################################################################################
|
|
175 if [[ -s errorFile ]]; then
|
|
176 #When the option --n is enabled, alfa prints '### End of the program' in stderr even if the process worked-
|
|
177 #The following lines to avoid the tool from crashing in this case
|
|
178 endProgram=`grep -c '### End of program' errorFile`
|
|
179 if [ "$endProgram" == "0" ]; then
|
|
180 #When alfa prints '### End of program' in stdout, all the messages in stderr are considered
|
|
181 #as warnings and not as errors. True errors make the script exits with code "2"
|
|
182 endProgram=`grep -c '### End of program' $logReport`
|
|
183 if [ "$endProgram" == "0" ]; then
|
|
184 >&2 printf "The script ALFA.py encountered the following error:\n\n"
|
|
185 >&2 cat errorFile
|
|
186 printf "ALFA error:\n" >> $logReport
|
|
187 cat errorFile >> $logReport
|
|
188 printf "\n******************************************************************\n" >> $logReport
|
|
189 exit 2
|
|
190 else
|
|
191 >&2 printf "The script ALFA.py encountered the following warning:\n\n"
|
|
192 >&2 cat errorFile
|
|
193 printf "ALFA warning:\n" >> $logReport
|
|
194 cat errorFile >> $logReport
|
|
195 printf "\n******************************************************************\n" >> $logReport
|
|
196 fi
|
|
197 fi
|
|
198 fi
|
|
199
|
|
200 #########################################################################################################
|
|
201 # OUTPUT REDIRECTIONS #
|
|
202 #########################################################################################################
|
|
203 if [ "$plotChoice" == "True" ]; then
|
|
204 if [ "$plotFormat" == "pdf" ]; then
|
|
205 mv "plotFile.pdf" $outputPdf;
|
|
206 elif [ "$plotFormat" == "png" ]; then
|
|
207 mv "plotFile.categories.png" $outputCategoriesPng;
|
|
208 mv "plotFile.biotypes.png" $outputBiotypesPng;
|
|
209 else
|
|
210 mv "plotFile.categories.svg" $outputCategoriesSvg;
|
|
211 mv "plotFile.biotypes.svg" $outputBiotypesSvg;
|
|
212 fi
|
|
213 fi
|
|
214 if [ "$countFileChoice" == "True" ]; then
|
|
215 > countFile;
|
|
216 for (( i = 1; i <= readsListLen; i++ )) do
|
|
217 printf "##LABEL: "${readsLabel[$i]}"\n\n" >> countFile;
|
|
218 cat ${readsLabel[$i]}".categories_counts" >> countFile;
|
|
219 printf "__________________________________________________________________\n" >> countFile;
|
|
220 done
|
|
221 mv countFile $outputCountFile;
|
|
222 fi
|
|
223 if [ "$indexChoice" == "True" ]; then
|
|
224 if [ "$annotationSource" == "index" ]; then
|
|
225 mv $strandedIndex $outputStrandedIndex
|
|
226 mv $unstrandedIndex $outputUnstrandedIndex
|
|
227 elif [ "$annotationSource" == "built_in_index" ]; then
|
|
228 cp $built_in_index_prefix".stranded.index" $outputStrandedIndex
|
|
229 cp $built_in_index_prefix".unstranded.index" $outputUnstrandedIndex
|
|
230 else
|
|
231 annotationFileName=`grep -P -o '[^/]*\.dat$' <<< $annotationFile`
|
|
232 mv $annotationFileName".stranded.index" $outputStrandedIndex
|
|
233 mv $annotationFileName".unstranded.index" $outputUnstrandedIndex
|
|
234 fi
|
|
235 fi |