Mercurial > repos > charles-bernard > alfa
comparison ALFA/alfa_wrapper.sh @ 18:a1e2ab10b317 draft
Uploaded
| author | charles-bernard |
|---|---|
| date | Tue, 11 Oct 2016 09:18:48 -0400 |
| parents | |
| children | 1239f24962d8 |
comparison
equal
deleted
inserted
replaced
| 17:e3d439570972 | 18:a1e2ab10b317 |
|---|---|
| 1 #!usr/bin/bash | |
| 2 | |
| 3 ######################################################################################################### | |
| 4 # ARGUMENTS FROM alfa_wrapper.xml # | |
| 5 ######################################################################################################### | |
| 6 configFile=$1; | |
| 7 logReport=$2; | |
| 8 sed -i -e '/^$/d; s/\t//g;' $configFile; | |
| 9 printf "__________________________________________________________________\n\n" > $logReport | |
| 10 printf " ALFA CONFIG \n" >> $logReport | |
| 11 printf "__________________________________________________________________\n" >> $logReport | |
| 12 cat $configFile >> $logReport | |
| 13 | |
| 14 ######################################################################################################### | |
| 15 # INITIALIZATION OF THE VARIABLES from $configFile # | |
| 16 ######################################################################################################### | |
| 17 #_INPUT1 | |
| 18 annotationSource=`grep -P '^annotationSource ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 19 if [ "$annotationSource" == "personal_gtf" ]; then | |
| 20 annotationFile=`grep -P '^annotationFile ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 21 elif [ "$annotationSource" == "built_in_index" ]; then | |
| 22 built_in_index_prefix=`grep -P '^built_in_index_prefix ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 23 else | |
| 24 strandedIndex=`grep -P '^strandedIndex ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 25 unstrandedIndex=`grep -P '^unstrandedIndex ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 26 fi | |
| 27 | |
| 28 #_INPUT2 | |
| 29 readsType=`grep -P '^readsType ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 30 readsFileList=`grep -P '^readsFile\[[0-9]+\] ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 31 readsLabelList=`grep -P '^readsLabel\[[0-9]+\] ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 32 | |
| 33 #_OUTPUT CHOICES | |
| 34 plotChoice=`grep -P '^plotChoice ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 35 countFileChoice=`grep -P '^countFileChoice ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 36 indexChoice=`grep -P '^indexChoice ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 37 | |
| 38 #_OUTPUT OPTIONS | |
| 39 strandness=`grep -P '^strandness ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 40 categoriesDepth=`grep -P '^categoriesDepth ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 41 plotFormat=`grep -P '^plotFormat ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 42 plotThresholdChoice=`grep -P '^plotThresholdChoice ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 43 if [ "$plotThresholdChoice" == "True" ]; then | |
| 44 yMin=`grep -P '^yMin ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 45 yMax=`grep -P '^yMax ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 46 fi | |
| 47 | |
| 48 #_OUTPUT FILES | |
| 49 if [ "$plotChoice" == "True" ]; then | |
| 50 if [ "$plotFormat" == "pdf" ]; then | |
| 51 outputPdf=`grep -P '^outputPdf ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 52 elif [ "$plotFormat" == "svg" ]; then | |
| 53 outputCategoriesSvg=`grep -P '^outputCategoriesSvg ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 54 outputBiotypesSvg=`grep -P '^outputBiotypesSvg ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 55 else | |
| 56 outputCategoriesPng=`grep -P '^outputCategoriesPng ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 57 outputBiotypesPng=`grep -P '^outputBiotypesPng ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 58 fi | |
| 59 fi | |
| 60 if [ "$countFileChoice" == "True" ]; then | |
| 61 outputCountFile=`grep -P '^outputCountFile ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 62 fi | |
| 63 if [ "$indexChoice" == "True" ]; then | |
| 64 outputStrandedIndex=`grep -P '^outputStrandedIndex ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 65 outputUnstrandedIndex=`grep -P '^outputUnstrandedIndex ?=' $configFile | awk 'BEGIN{FS="= ?"} {print $2}'`; | |
| 66 fi | |
| 67 | |
| 68 ######################################################################################################### | |
| 69 # CREATION OF A TMP DIRECTORY FOR THE OUTPUT FILES OF ALFA AND cd # | |
| 70 ######################################################################################################### | |
| 71 outputDirectory=`mktemp -d /export/home1/users/biocomp/chbernar/galaxy/database/tmp/tmpXXXXXX`; | |
| 72 if [ -d $outputDirectory ]; then | |
| 73 chmod -R ugo+wrx $outputDirectory; | |
| 74 rm -R $outputDirectory; | |
| 75 fi | |
| 76 mkdir $outputDirectory; | |
| 77 chmod -R ugo+wrx $outputDirectory; | |
| 78 cd $outputDirectory; | |
| 79 | |
| 80 ######################################################################################################### | |
| 81 # TEST OF INPUT1 # | |
| 82 ######################################################################################################### | |
| 83 if [ "$annotationSource" == "index" ]; then | |
| 84 #need to copy the files.dat to .*index because ALFA requires the extension ".(un)stranded.index" | |
| 85 index="index" | |
| 86 cp $strandedIndex $index".stranded.index" | |
| 87 cp $unstrandedIndex $index".unstranded.index" | |
| 88 fi | |
| 89 | |
| 90 ######################################################################################################### | |
| 91 # TEST OF INPUT2 AND DETERMINATION OF PYTHON READS INPUT ARGUMENT # | |
| 92 ######################################################################################################### | |
| 93 readsListLen=`echo "$readsFileList" | wc -l`; | |
| 94 readsInput=""; | |
| 95 for (( i = 1; i <= readsListLen; i++ )) do | |
| 96 readsFile[$i]=`echo "$readsFileList" | awk -v i=$i 'NR==i'`; | |
| 97 readsLabel[$i]=`echo "$readsLabelList" | awk -v i=$i 'NR==i' | sed -e 's/ /_/g'`; | |
| 98 if [ "$readsType" == "bam" ]; then | |
| 99 bamSorted=`samtools view -H "${readsFile[$i]}" | grep -c 'SO:unsorted'` | |
| 100 if [ "$bamSorted" != "0" ] ; then | |
| 101 samtools sort ${readsFile[$i]} ${readsFile[$i]} | |
| 102 fi | |
| 103 else | |
| 104 #need to copy the file.dat to tmp.bedgraph because ALFA requires the extension ".bedgraph" | |
| 105 bedgraphFile="tmpBedgraph_"$i | |
| 106 cp ${readsFile[$i]} $bedgraphFile".bedgraph" | |
| 107 readsFile[$i]=$bedgraphFile | |
| 108 fi | |
| 109 if [ "${readsLabel[$i]}" == "" ]; then | |
| 110 readsLabel[$i]="sample_""$i"; | |
| 111 fi | |
| 112 readsInput=$readsInput" "${readsFile[$i]}" "${readsLabel[$i]}; | |
| 113 done | |
| 114 | |
| 115 ######################################################################################################### | |
| 116 # DETERMINATION OF THE APPROPRIATE SCRIPTS ARGUMENTS # | |
| 117 ######################################################################################################### | |
| 118 scriptPath="/export/home1/users/biocomp/chbernar/galaxy/tools/alfa/"; | |
| 119 if [ "$annotationSource" == "index" ]; then | |
| 120 scriptInput="-g $index -i ""$readsInput"; | |
| 121 elif [ "$annotationSource" == "built_in_index" ]; then | |
| 122 scriptInput="-g $built_in_index_prefix -i ""$readsInput"; | |
| 123 else | |
| 124 scriptInput="-a $annotationFile -i ""$readsInput"; | |
| 125 fi | |
| 126 if [ "$readsType" = "bedgraph" ]; then | |
| 127 scriptInput=$scriptInput" --bedgraph"; | |
| 128 fi | |
| 129 scriptStrandness="-s "$strandness | |
| 130 scriptCategoriesDepth="-d "$categoriesDepth | |
| 131 if [ "$plotChoice" == "True" ]; then | |
| 132 if [ "$plotFormat" == "pdf" ]; then | |
| 133 scriptPlotOutput="--pdf plotFile.pdf"; | |
| 134 else | |
| 135 scriptPlotOutput="--"$plotFormat" plotFile"; | |
| 136 fi | |
| 137 if [ "$plotThresholdChoice" == "True" ]; then | |
| 138 scriptPlotOutput=$scriptPlotOutput" -t ""$yMin"" ""$yMax" | |
| 139 fi | |
| 140 else | |
| 141 scriptPlotOutput="--n"; | |
| 142 fi | |
| 143 | |
| 144 ######################################################################################################### | |
| 145 # DISPLAY ALFA PROCESS # | |
| 146 ######################################################################################################### | |
| 147 printf "__________________________________________________________________\n\n" >> $logReport | |
| 148 printf " ALFA PROCESS \n" >> $logReport | |
| 149 printf "__________________________________________________________________\n" >> $logReport | |
| 150 | |
| 151 if [ "$plotChoice" == "False" ] && [ "$countFileChoice" == "False" ] && [ "$indexChoice" == "False" ]; then | |
| 152 cat <<error 1>&2 | |
| 153 | |
| 154 No output to return. | |
| 155 Process Aborted | |
| 156 error | |
| 157 exit 0 | |
| 158 fi | |
| 159 | |
| 160 printf "Command:\n" >> $logReport | |
| 161 echo "python ""$scriptPath"ALFA.py $scriptInput $scriptStrandness $scriptCategoriesDepth $scriptPlotOutput >> $logReport; | |
| 162 printf "\n******************************************************************\n" >> $logReport | |
| 163 printf "Temporary Output Directory:\n" >> $logReport | |
| 164 echo $outputDirectory >> $logReport | |
| 165 printf "\n******************************************************************\n" >> $logReport | |
| 166 printf "ALFA prompt:\n" >> $logReport | |
| 167 python "$scriptPath"ALFA.py $scriptInput $scriptStrandness $scriptCategoriesDepth $scriptPlotOutput >> $logReport 2>errorFile; | |
| 168 printf "\n******************************************************************\n" >> $logReport | |
| 169 | |
| 170 ######################################################################################################### | |
| 171 # REDIRECTION OF ERRORS - TMP SOURCE ALFA.PY MUST BE CORRECTED SOON # | |
| 172 ######################################################################################################### | |
| 173 if [[ -s errorFile ]]; then | |
| 174 #When the option --n is enabled, alfa prints '### End of the program' in stderr even if the process worked- | |
| 175 #The following lines to avoid the tool from crashing in this case | |
| 176 endProgram=`grep -c '### End of program' errorFile` | |
| 177 if [ "$endProgram" == "0" ]; then | |
| 178 #When alfa prints '### End of program' in stdout, all the messages in stderr are considered | |
| 179 #as warnings and not as errors. True errors make the script exits with code "2" | |
| 180 endProgram=`grep -c '### End of program' $logReport` | |
| 181 if [ "$endProgram" == "0" ]; then | |
| 182 >&2 printf "The script ALFA.py encountered the following error:\n\n" | |
| 183 >&2 cat errorFile | |
| 184 printf "ALFA error:\n" >> $logReport | |
| 185 cat errorFile >> $logReport | |
| 186 printf "\n******************************************************************\n" >> $logReport | |
| 187 exit 2 | |
| 188 else | |
| 189 >&2 printf "The script ALFA.py encountered the following warning:\n\n" | |
| 190 >&2 cat errorFile | |
| 191 printf "ALFA warning:\n" >> $logReport | |
| 192 cat errorFile >> $logReport | |
| 193 printf "\n******************************************************************\n" >> $logReport | |
| 194 fi | |
| 195 fi | |
| 196 fi | |
| 197 | |
| 198 ######################################################################################################### | |
| 199 # OUTPUT REDIRECTIONS # | |
| 200 ######################################################################################################### | |
| 201 if [ "$plotChoice" == "True" ]; then | |
| 202 if [ "$plotFormat" == "pdf" ]; then | |
| 203 mv "plotFile.pdf" $outputPdf; | |
| 204 elif [ "$plotFormat" == "png" ]; then | |
| 205 mv "plotFile.categories.png" $outputCategoriesPng; | |
| 206 mv "plotFile.biotypes.png" $outputBiotypesPng; | |
| 207 else | |
| 208 mv "plotFile.categories.svg" $outputCategoriesSvg; | |
| 209 mv "plotFile.biotypes.svg" $outputBiotypesSvg; | |
| 210 fi | |
| 211 fi | |
| 212 if [ "$countFileChoice" == "True" ]; then | |
| 213 > countFile; | |
| 214 for (( i = 1; i <= readsListLen; i++ )) do | |
| 215 printf "##LABEL: "${readsLabel[$i]}"\n\n" >> countFile; | |
| 216 cat ${readsLabel[$i]}".categories_counts" >> countFile; | |
| 217 printf "__________________________________________________________________\n" >> countFile; | |
| 218 done | |
| 219 mv countFile $outputCountFile; | |
| 220 fi | |
| 221 if [ "$indexChoice" == "True" ]; then | |
| 222 if [ "$annotationSource" == "index" ]; then | |
| 223 mv $strandedIndex $outputStrandedIndex | |
| 224 mv $unstrandedIndex $outputUnstrandedIndex | |
| 225 elif [ "$annotationSource" == "built_in_index" ]; then | |
| 226 cp $built_in_index_prefix".stranded.index" $outputStrandedIndex | |
| 227 cp $built_in_index_prefix".unstranded.index" $outputUnstrandedIndex | |
| 228 else | |
| 229 annotationFileName=`grep -P -o '[^/]*\.dat$' <<< $annotationFile` | |
| 230 mv $annotationFileName".stranded.index" $outputStrandedIndex | |
| 231 mv $annotationFileName".unstranded.index" $outputUnstrandedIndex | |
| 232 fi | |
| 233 fi |