Implements #39 , which have the side effect to disable weblogo comput…

…ation as no info-file can be outputed by cutadapt while using the linked adapter strategy

bin/make_doc.sh

@@ -87,7 +87,7 @@ then
 	FWDSIM=`awk -v F=${#FWD} -v D=$PDIFFS 'BEGIN{printf "%.2g\n", 1-D/F}'`
 	RVSSIM=`awk -v R=${#RVS} -v D=$PDIFFS 'BEGIN{printf "%.2g\n", 1-D/R}'`
 	echo "Read pairs were extracted from raw libraries if at least one of the two reads hold the expected primer (forward primer for forward library, reverse primer for reverse library) at its 5' end, with a similarity threshold of $FWDSIM and $RVSSIM for the forward and reverse primer, respectively."
-	CUTAVG=`grep -m 1 "Total read pairs processed:" libraries/fastq/log.cutadapt.* | awk '{sub(",","",$NF);sum+=$NF}END{printf "%.0f\n", sum/NR}'`
+	CUTAVG=`grep -m 1 "Total read pairs processed:" libraries/fastq/log.cutadapt.* | awk '{gsub(",","",$NF);sum+=$NF}END{printf "%.0f\n", sum/NR}'`
 	echo "An average of $CUTAVG reads was extracted per pair of libraries."
 	echo ""
 fi

src/Illumina_fastq.step

@@ -14,16 +14,20 @@ cd libraries
 # Primer clipping
 if [ $CLIPPING == "yes" ]
 then
-	#FWDDIS=`awk -v F=${#FWD} -v D=$PDIFFS 'BEGIN{printf "%.2g\n", D/F}'`
-	RVSDIS=`awk -v R=${#RVS} -v D=$PDIFFS 'BEGIN{printf "%.2g\n", D/R}'`
+	FWD_RC=$(echo ">fwd#$FWD" | tr "#" "\n" | vsearch --quiet --fastx_revcomp - --fastaout - | grep -v ">")
+	RVS_RC=$(echo ">rvs#$RVS" | tr "#" "\n" | vsearch --quiet --fastx_revcomp - --fastaout - | grep -v ">")
 	DISS=`awk -v F=${#FWD} -v R=${#RVS} -v D=$PDIFFS 'BEGIN{DISS=D/F;if(D/R>DISS){DISS=D/R};printf "%.2g\n", DISS}'`
 	# cutadapt
-	cutadapt -g $FWD -G $RVS -e $DISS --trimmed-only --no-indels -o fastq/$LIB_NAME.fwd.fastq -p fastq/$LIB_NAME.rvs.fastq $FWD_LIB.fastq $RVS_LIB.fastq --info-file=raw_stat/$FWD_LIB.cutadapt > fastq/log.cutadapt.$LIB_NAME.txt
-	cutadapt -g $RVS -O ${#RVS} -e $RVSDIS --no-indels --trimmed-only -o fastq/$LIB_NAME.rvs.tmp.fastq --info-file=raw_stat/$RVS_LIB.cutadapt $RVS_LIB.fastq > fastq/log.cutadapt.aux.$LIB_NAME.txt
+	# in forward primer direction (and conversely for reverse library)
+	cutadapt -a $FWD...$RVS_RC -A $RVS...$FWD_RC -e $DISS --no-indels --untrimmed-output=fastq/$FWD_LIB.fwd_unt.fastq --untrimmed-paired-output=fastq/$RVS_LIB.fwd_unt.fastq -o fastq/$FWD_LIB.fwd.fastq -p fastq/$RVS_LIB.fwd.fastq $FWD_LIB.fastq $RVS_LIB.fastq > fastq/log.cutadapt.$LIB_NAME.fwd.txt
+	# in reverse primer direction and reverse complement
+	cutadapt -a $RVS...$FWD_RC -A $FWD...$RVS_RC -e $DISS --no-indels --trimmed-only -o fastq/$FWD_LIB.rvs.fastq -p fastq/$RVS_LIB.rvs.fastq $FWD_LIB.fwd_unt.fastq $RVS_LIB.fwd_unt.fastq > fastq/log.cutadapt.$LIB_NAME.rvs.txt
+	vsearch --quiet --fastx_revcomp fastq/$FWD_LIB.rvs.fastq --fastqout - | cat fastq/$FWD_LIB.fwd.fastq - > fastq/$LIB_NAME.fwd.fastq
+	vsearch --quiet --fastx_revcomp fastq/$RVS_LIB.rvs.fastq --fastqout - | cat fastq/$RVS_LIB.fwd.fastq - > fastq/$LIB_NAME.rvs.fastq
+	rm fastq/$FWD_LIB.rvs.fastq fastq/$FWD_LIB.fwd.fastq fastq/$FWD_LIB.fwd_unt.fastq fastq/$RVS_LIB.rvs.fastq fastq/$RVS_LIB.fwd.fastq fastq/$RVS_LIB.fwd_unt.fastq
 	# primer logo
-	weblogo -c classic -s large -t "$LIB_NAME: $FWD_NAME" < <(awk -v a=${#FWD} 'BEGIN{FS="\t"} NF==11{printf ">%s\n%*s\n",$1,a,$6}' raw_stat/$FWD_LIB.cutadapt | sed 's/ /N/g' )> raw_stat/weblogo.$LIB_NAME.forward.eps
-	weblogo -c classic -s large -t "$LIB_NAME: primer $RVS_NAME" < <(awk -v a=${#RVS} 'BEGIN{FS="\t"} NF==11{printf ">%s\n%*s\n",$1,a,$6}' raw_stat/$RVS_LIB.cutadapt | sed 's/ /N/g') > raw_stat/weblogo.$LIB_NAME.reverse.eps
-	rm fastq/$LIB_NAME.rvs.tmp.fastq
+	# weblogo -c classic -s large -t "$LIB_NAME: $FWD_NAME" < <(awk -v a=${#FWD} 'BEGIN{FS="\t"} NF==11{printf ">%s\n%*s\n",$1,a,$6}' raw_stat/$FWD_LIB.cutadapt | sed 's/ /N/g' )> raw_stat/weblogo.$LIB_NAME.forward.eps
+	# weblogo -c classic -s large -t "$LIB_NAME: primer $RVS_NAME" < <(awk -v a=${#RVS} 'BEGIN{FS="\t"} NF==11{printf ">%s\n%*s\n",$1,a,$6}' raw_stat/$RVS_LIB.cutadapt | sed 's/ /N/g') > raw_stat/weblogo.$LIB_NAME.reverse.eps
 else
 	ln -s $PWD/$FWD_LIB.fastq $PWD/fastq/$LIB_NAME.fwd.fastq
 	ln -s $PWD/$RVS_LIB.fastq $PWD/fastq/$LIB_NAME.rvs.fastq

src/Illumina_quality.step

@@ -49,11 +49,11 @@ dd if=$SUBPROJECT.summary.stat.tsv of=../archives/$SUBPROJECT.outputs/$SUBPROJEC
 # Merge raw stat statistics and weblogos
 cd ..
 gs -q -sDEVICE=pdfwrite -o archives/$SUBPROJECT.outputs/$SUBPROJECT.raw_and_pair-end_reads_statistics.pdf libraries/raw_stat/$SUBPROJECT.raw_reads_with_primer_quality.pdf libraries/raw_stat/$SUBPROJECT.pair-end_reads_quality.pdf libraries/raw_stat/$SUBPROJECT.legend.pdf
-if [ $CLIPPING == "yes" ]
-then
-gs -q -sDEVICE=pdfwrite -dEPSCrop -o archives/$SUBPROJECT.outputs/$SUBPROJECT.forward.weblogo.pdf libraries/raw_stat/weblogo.*.forward.eps
-gs -q -sDEVICE=pdfwrite -dEPSCrop -o archives/$SUBPROJECT.outputs/$SUBPROJECT.reverse.weblogo.pdf libraries/raw_stat/weblogo.*.reverse.eps
-fi
+#if [ $CLIPPING == "yes" ]
+#then
+#gs -q -sDEVICE=pdfwrite -dEPSCrop -o archives/$SUBPROJECT.outputs/$SUBPROJECT.forward.weblogo.pdf libraries/raw_stat/weblogo.*.forward.eps
+#gs -q -sDEVICE=pdfwrite -dEPSCrop -o archives/$SUBPROJECT.outputs/$SUBPROJECT.reverse.weblogo.pdf libraries/raw_stat/weblogo.*.reverse.eps
+#fi
 
 # list files and directories
 . $BIN/list_step_files.sh