# HG changeset patch # User jbrayet # Date 1446128706 14400 # Node ID 1027a330d606abcf5db1f1bba78d37754f37e687 # Parent 0c34e0bef7d33e8ec94c2e6bd380c3f61c117ca3 Uploaded diff -r 0c34e0bef7d3 -r 1027a330d606 ncPRO-QC.sh --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/ncPRO-QC.sh Thu Oct 29 10:25:06 2015 -0400 @@ -0,0 +1,423 @@ +#!/bin/bash + +# I added the 'me' lines +#assumption : a user can't begin with a fastq file already grouped ! + +while getopts "i:s:n:g:m:o:f:r:h:p:l:t:a:" optionName; do +case "$optionName" in + +i) INPUT+="$OPTARG,";; +s) SAMPLENAME+="$OPTARG,";; +n) PROJECTNAME="$OPTARG";; +g) GENOME="$OPTARG";; +m) RMSK="$OPTARG";; +o) BAM_OUT+="$OPTARG,";; +f) RFAM="$OPTARG";; +r) REPORT="$OPTARG";; +h) HTML_REPORT="$OPTARG";; +p) PDF_REPORT="$OPTARG";; +l) LOG_FILE="$OPTARG";; +t) INPUT_TYPE="$OPTARG";; +a) ALIGNMENT="$OPTARG";; + +esac +done + +sampleArray=(${INPUT//,/ }) +nameArray=(${SAMPLENAME//,/ }) +bamArray=(${BAM_OUT//,/ }) + +if [[ $REPORT == "pdf" ]];then + OUTPUT_PATH_DIR=`dirname $LOG_FILE` + OUTPUT_PATH_NAME=`basename $LOG_FILE .dat` +else + OUTPUT_PATH_DIR=`dirname $HTML_REPORT` + OUTPUT_PATH_NAME=`basename $HTML_REPORT .dat` +fi + +OUTPUT_PATH="${OUTPUT_PATH_DIR}/${OUTPUT_PATH_NAME}_files" + + + +#ME +mkdir -p $OUTPUT_PATH + +chmod 777 -R $OUTPUT_PATH + +#VERSION=`echo $OUTPUT_PATH | cut -d"/" -f3` +#VERSION=`echo $VERSION | cut -d"_" -f2` + +#DEBUG_MODE + +DEBUG_MODE="on" +DEBUG="/dev/null" + +if [[ $DEBUG_MODE == "on" ]];then + + DEBUG="$OUTPUT_PATH/ncPRO-QC.debug" + +fi + +#Deploy ncPRO directories structure + +/bioinfo/local/curie/ngs-data-analysis/ncPRO-seq/bin/ncPRO-deploy -o $OUTPUT_PATH > $DEBUG + +echo "$INPUT" >> $DEBUG +echo "$SAMPLENAME" >> $DEBUG +echo "$PROJECTNAME" >> $DEBUG +echo "$GENOME" >> $DEBUG +echo "$RMSK" >> $DEBUG +echo "$BAM_OUT" >> $DEBUG +echo "$RFAM" >> $DEBUG +echo "$REPORT" >> $DEBUG +echo "$HTML_REPORT" >> $DEBUG +echo "$PDF_REPORT" >> $DEBUG +echo "$LOG_FILE" >> $DEBUG +echo "$INPUT_TYPE" >> $DEBUG +echo "$ALIGNMENT" >> $DEBUG + +echo "$sampleArray" >> $DEBUG +echo "$nameArray" >> $DEBUG +echo "$bamArray" >> $DEBUG + +#Go to working directory + +cd $OUTPUT_PATH + +rm annotation + +ln -s /bioinfo/local/curie/ngs-data-analysis/annotation . + +rm manuals + +#Create symbolic link to input +#********************************************************************************* NEW: for BAM files, check if reads in input are grouped or not and change cmd accordignly *********** +if [[ $INPUT_TYPE == "fastq" ]];then + + count=0 + for i in ${sampleArray[*]} + do + ln -s $i ${OUTPUT_PATH}/rawdata/${nameArray[count]}.fastq + count=$(( $count + 1 )) + done +fi + +if [[ $INPUT_TYPE == "bam" ]];then + + count=0 + for i in ${sampleArray[*]} + do + ln -s $i ${OUTPUT_PATH}/rawdata/${nameArray[count]}.bam + count=$(( $count + 1 )) + + #check if grouped + RG=`samtools view $i | awk --posix 'BEGIN {RG=1} { if ($1 !~ /^[0-9]{1,}_[0-9]{1,}$/) {RG=0 ; exit} } END { print RG}'` + done +fi + +#Edit config-ncrna.txt +CONFIG_FILE=config-ncrna.txt +sed -i "s/mm9/$GENOME/g" $CONFIG_FILE +sed -i "s/hg19/$GENOME/g" $CONFIG_FILE + +sed -i "/N_CPU/c\N_CPU = 6" $CONFIG_FILE #****** Make sure this value matches universe.ini files + +sed -i "s/test_Curie/$PROJECTNAME/g" $CONFIG_FILE + + +#Build command line + +if [[ $INPUT_TYPE == "fastq" ]];then + + if [[ $ALIGNMENT == "True" ]]; then + + + COMMAND_LINE="-c $CONFIG_FILE -s processRead -s mapGenome -s mapGenomeStat -s mapAnnOverview" + + if [[ $RFAM == "True" ]];then + + COMMAND_LINE="$COMMAND_LINE -s overviewRfam -s overviewRmsk" + + fi + + else + + COMMAND_LINE="-c $CONFIG_FILE -s processRead" + + fi + +fi + + + +#### NEW if BAM already grouped, omit [ -s processBam ] + put input.bam in /bowtie_results +if [[ $INPUT_TYPE == "bam" ]];then + + if [[ $RG = 0 ]]; then #if bam file is NOT grouped + + COMMAND_LINE="-c $CONFIG_FILE -s processBam -s mapGenomeStat -s mapAnnOverview" + + else + + count=0 + for i in ${sampleArray[*]} + do + ln -s $i ${OUTPUT_PATH}/rawdata/${nameArray[count]}.bam + count=$(( $count + 1 )) + done + COMMAND_LINE="-c $CONFIG_FILE -s mapGenomeStat -s mapAnnOverview" + fi + + + if [[ $RFAM == "True" ]];then + + COMMAND_LINE="$COMMAND_LINE -s overviewRfam -s overviewRmsk" + + fi + +fi + +#************************* new + +##### Function to create HTML report in Galaxy ###### + +function createHtmlReport +{ + + # galaxy part : + + #Reformat html output + + tr '>' '\n' < ${OUTPUT_PATH}/report.html | sed -ne "s@.* ${OUTPUT_PATH}/img_list.txt + + NC_LOGO=`head -1 ${OUTPUT_PATH}/img_list.txt` + + #HTTP_PATH=`echo $OUTPUT_PATH | sed "s/\/data\/kdi_${VERSION}/http:\/\/data-kdi-${VERSION}.curie.fr\/file/g"` + + i=1 + + if [[ $INPUT_TYPE == "fastq" ]];then + + echo "

QUALITY CONTROL

" > ${HTML_REPORT} + fi + + if [[ $INPUT_TYPE == "bam" ]];then + + echo "

DATA MAPPING

" > $HTML_REPORT + fi + + while read line + do + + if [[ $line != "" ]];then + + if [[ $i == "3" ]];then + + ahref=`sed "s/html\/thumb/pic/" <<< $line` + + if [[ $INPUT_TYPE == "fastq" ]];then + echo "

Base Composition Information

" >> $HTML_REPORT + fi + + if [[ $INPUT_TYPE == "bam" ]];then + echo "

Mapping proportions

" >> $HTML_REPORT + fi + + fi + + if [[ $i == "4" ]];then + + ahref=`sed "s/html\/thumb/pic/" <<< $line` + + if [[ $INPUT_TYPE == "fastq" ]];then + echo "
" >> $HTML_REPORT + fi + + if [[ $INPUT_TYPE == "bam" ]];then + echo "

Distinct Reads Length Distribution

" >> $HTML_REPORT + fi + + fi + + if [[ $i == "5" ]];then + + ahref=`sed "s/html\/thumb/pic/" <<< $line` + + if [[ $INPUT_TYPE == "fastq" ]];then + echo "

Distinct Reads Length Distribution

" >> $HTML_REPORT + fi + + if [[ $INPUT_TYPE == "bam" ]];then + echo "

Abundant Reads Length Distribution

" >> $HTML_REPORT + fi + + fi + + if [[ $i == "6" ]];then + + ahref=`sed "s/html\/thumb/pic/" <<< $line` + + if [[ $INPUT_TYPE == "fastq" ]];then + echo "

Quality Score

" >> $HTML_REPORT + fi + + if [[ $INPUT_TYPE == "bam" ]];then + echo "

ncRNAs OVERVIEW

Reads Annotation Overview

" >> $HTML_REPORT + fi + + fi + + if [[ $i == "7" ]];then + + ahref=`sed "s/html\/thumb/pic/" <<< $line` + if [[ $INPUT_TYPE == "fastq" ]];then + echo "

Abundant Reads Length Distribution

" >> $HTML_REPORT + fi + if [[ $INPUT_TYPE == "bam" ]];then + + if [[ $RFAM == "True" ]];then + echo "

Precursor miRNAs Annotation

" >> $HTML_REPORT + else + break + fi + fi + fi + + if [[ $i == "8" ]];then + + ahref=`sed "s/html\/thumb/pic/" <<< $line` + if [[ $INPUT_TYPE == "fastq" ]];then + + if [[ $ALIGNMENT == "True" ]]; then + + echo "

DATA MAPPING

Mapping proportions

" >> $HTML_REPORT + else + break + fi + fi + if [[ $INPUT_TYPE == "bam" ]];then + + echo "

Annotation of ncRNAs from RFAM

" >> $HTML_REPORT + fi + fi + + if [[ $i == "9" ]];then + + ahref=`sed "s/html\/thumb/pic/" <<< $line` + if [[ $INPUT_TYPE == "fastq" ]];then + echo "

Distinct Reads Length Distribution

" >> $HTML_REPORT + fi + + if [[ $INPUT_TYPE == "bam" ]];then + + echo "

Annotation of Repetitive Regions

" >> $HTML_REPORT + break + fi + + fi + + if [[ $i == "10" ]];then + + ahref=`sed "s/html\/thumb/pic/" <<< $line` + + echo "

Abundant Reads Length Distribution

" >> $HTML_REPORT + + fi + + if [[ $i == "11" ]];then + + ahref=`sed "s/html\/thumb/pic/" <<< $line` + + echo "

ncRNAs OVERVIEW

Reads Annotation Overview

" >> $HTML_REPORT + + fi + + if [[ $RFAM == "True" ]];then + + if [[ $i == "12" ]];then + + ahref=`sed "s/html\/thumb/pic/" <<< $line` + + echo "

Precursor miRNAs Annotation

" >> $HTML_REPORT + + fi + + if [[ $i == "13" ]];then + + ahref=`sed "s/html\/thumb/pic/" <<< $line` + + echo "

Annotation of ncRNAs from RFAM

" >> $HTML_REPORT + + fi + + if [[ $i == "14" ]];then + + ahref=`sed "s/html\/thumb/pic/" <<< $line` + + echo "

Annotation of Repetitive Regions

" >> $HTML_REPORT + + fi + + fi + + i=$(( $i + 1 )) + + fi + + + done < ${OUTPUT_PATH}/img_list.txt + + rm ${OUTPUT_PATH}/img_list.txt + + + +} + + +#Launch ncPRO analysis + +#FAIRE une boucle pour -s hrml_builder ou -s pdf_builder + +if [[ $REPORT == "all" ]];then + + + /bioinfo/local/curie/ngs-data-analysis/ncPRO-seq/bin/ncPRO-seq $COMMAND_LINE -s html_builder -s pdf_builder>> $DEBUG + + createHtmlReport + + cp ${OUTPUT_PATH}/Analysis_report_ncPRO-seq.pdf $PDF_REPORT + +fi + +if [[ $REPORT == "pdf" ]];then + + + /bioinfo/local/curie/ngs-data-analysis/ncPRO-seq/bin/ncPRO-seq $COMMAND_LINE -s pdf_builder>> $DEBUG + + cp ${OUTPUT_PATH}/Analysis_report_ncPRO-seq.pdf $PDF_REPORT + + +fi + +if [[ $REPORT == "html" ]];then + + + /bioinfo/local/curie/ngs-data-analysis/ncPRO-seq/bin/ncPRO-seq $COMMAND_LINE -s html_builder>> $DEBUG + + createHtmlReport + +fi + +#Galaxy output handling + +cp ${OUTPUT_PATH}/pipeline.log $LOG_FILE + +if [[ $ALIGNMENT == "True" ]];then + + count=0 + for i in ${bamArray[*]} + do + cp ${OUTPUT_PATH}/bowtie_results/${nameArray[count]/_/.}_$GENOME.bam $i + count=$(( $count + 1 )) + done +fi