replace UMI with work (hardcoded for converted barcode length) (88591108) · Commits · github_fork / Universc

launch_universc.sh

+57 −42

Original line number	Original line	Diff line number	Diff line
	@@ -1657,7 +1657,7 @@ if [[ -n "$barcodefile" ]]; then
	barcodefile=$(readlink -f $barcodefile)		barcodefile=$(readlink -f $barcodefile)
	custombarcodes=true		custombarcodes=true
	#allowing WellList from ICELL8 and other well-based techniques		#allowing WellList from ICELL8 and other well-based techniques
	if [[ "$technology" == "bd-rhapsody" ]] \|\| [[ "$technology" == "icell8" ]] \|\| [[ "$technology" == "quartz-seq" ]] \|\| [[ "$technology" == "ramda-seq" ]] [[ "$technology" == "quartz-seq2" ]] \|\| [[ "$technology" == "microwellseq" ]] \|\| [[ "$technology" == "smartseq" ]] \|\| [[ "$technology" == "seqwell" ]] \|\| [[ "$technology" == "sciseq2" ]] \|\| [[ "$technology" == "sciseq3" ]] \|\| [[ "$technology" == "scifiseq" ]] \|\| [[ "$technology" == "splitseq" ]] \|\| [[ "$technology" == "splitseq2" ]] \|\| [[ "$technology" == "custom" ]]; then		if [[ "$technology" == "bd-rhapsody" ]] \|\| [[ "$technology" == "icell8" ]] \|\| [[ "$technology" == "quartz-seq" ]] \|\| [[ "$technology" == "ramda-seq" ]] \|\| [[ "$technology" == "quartz-seq2" ]] \|\| [[ "$technology" == "microwellseq" ]] \|\| [[ "$technology" == "smartseq" ]] \|\| [[ "$technology" == "seqwell" ]] \|\| [[ "$technology" == "sciseq2" ]] \|\| [[ "$technology" == "sciseq3" ]] \|\| [[ "$technology" == "scifiseq" ]] \|\| [[ "$technology" == "splitseq" ]] \|\| [[ "$technology" == "splitseq2" ]] \|\| [[ "$technology" == "custom" ]]; then
	seg=$'\t'		seg=$'\t'
	n_col=$(awk -F'\t' '{print NF}' $barcodefile \| sort -nu \| tail -n 1)		n_col=$(awk -F'\t' '{print NF}' $barcodefile \| sort -nu \| tail -n 1)
	if [[ $n_col -eq 1 ]]; then		if [[ $n_col -eq 1 ]]; then
	@@ -2627,7 +2627,7 @@ else
	perl sub/AddMockUMI.pl --fastq=${convR1} --out_dir $crIN --head_length=$barcodelength --umi_length=$umi_default		perl sub/AddMockUMI.pl --fastq=${convR1} --out_dir $crIN --head_length=$barcodelength --umi_length=$umi_default
	umilength=$umi_default		umilength=$umi_default
	umiadjust=0		umiadjust=0
	if [[ $chemistry == "SC3Pv3"]; then		if [[ $chemistry == "SC3Pv3" ]]; then
	chemistry="SC3Pv2"		chemistry="SC3Pv2"
	fi		fi

	@@ -2761,7 +2761,7 @@ else
	perl sub/AddMockUMI.pl --fastq=${convR1} --out_dir $crIN --head_length=$barcodelength --umi_length=$umi_default		perl sub/AddMockUMI.pl --fastq=${convR1} --out_dir $crIN --head_length=$barcodelength --umi_length=$umi_default
	umilength=$umi_default		umilength=$umi_default
	umiadjust=0		umiadjust=0
	if [[ $chemistry == "SC3Pv3"]; then		if [[ $chemistry == "SC3Pv3" ]]; then
	chemistry="SC3Pv2"		chemistry="SC3Pv2"
	fi		fi
	#returns a combined R1 file with barcode and mock UMI		#returns a combined R1 file with barcode and mock UMI
	@@ -3215,6 +3215,26 @@ else
	fi		fi
	fi		fi

			#convert UMI
			echo " adjusting UMIs of R1 files"
			# check if original UMI is shorter than default
			if [[ 0 -gt $umiadjust ]]; then
			for convFile in "${convFiles[@]}"; do
			echo " handling $convFile ..."
			toS=`printf '%0.sA' $(seq 1 $(($umiadjust * -1)))`
			toQ=`printf '%0.sI' $(seq 1 $(($umiadjust * -1)))`
			#compute length of adjusted barcode + original UMI
			keeplength=`echo $((${barcode_default}+${umi_default}-($umiadjust * -1)))`
			#Add n characters to the end of the sequence
			sed -E "2~4s/(.{$keeplength})(.*)/\1$toS\2/" $convFile > ${crIN}/.temp
			mv ${crIN}/.temp $convFile
			#Add n characters to the end of the quality
			sed -E "4~4s/(.{$keeplength})(.*)/\1$toQ\2/" $convFile > ${crIN}/.temp
			mv ${crIN}/.temp $convFile
			echo " ${convFile} adjusted"
			done
			fi

	#replace UMI with mock UMI to count reads (for technologies not already containing mock UMI)		#replace UMI with mock UMI to count reads (for technologies not already containing mock UMI)
	if [[ $technology != "icell8" ]] && [[ $technology != "ramda-seq" ]] && [[ $technology != "quartz-seq" ]] && [[ $technology != "smartseq" ]] && [[ $technology != "smartseq2" ]] && [[ $technology != "strt-seq" ]]; then		if [[ $technology != "icell8" ]] && [[ $technology != "ramda-seq" ]] && [[ $technology != "quartz-seq" ]] && [[ $technology != "smartseq" ]] && [[ $technology != "smartseq2" ]] && [[ $technology != "strt-seq" ]]; then
	if [[ $nonUMI ]]; then		if [[ $nonUMI ]]; then
	@@ -3222,56 +3242,51 @@ else
	echo "NOTICE: results will result read counts not UMI"		echo "NOTICE: results will result read counts not UMI"
	echo "## this behaviour is not recommended unless integrating with non-UMI data ##"		echo "## this behaviour is not recommended unless integrating with non-UMI data ##"

			for convFile in "${convFiles[@]}"; do
			convR1=$convFile
	#remove inflated umi (to replace with mock and count as reads)		#remove inflated umi (to replace with mock and count as reads)
	cmd=$(echo 'sed -E "		sed -E "
	/^(.{'$barcodelength'})(.{'${umilength}'})(.*)/ {		/^(.{16})(.{10})(.*)/ {
	s/^(.{'$barcodelength'})(.{'${umilength}'})(.*)/\1\3/g		s/^(.{16})(.{10})(.*)/\1\3/g
	n		n
	n		n
	s/^(.{'$barcodelength'})(.{'${umilength}'})(.*)/\1\3/g		s/^(.{16})(.{10})(.*)/\1\3/g
	}" $convFile > ${crIN}/.temp		}" $convFile > ${crIN}/.temp\n
	mv ${crIN}/.temp $convFile')		mv ${crIN}/.temp $convFile'
			if [[ $chemistry == "SC3Pv3" ]]; then
			chemistry="SC3Pv2"
			fi
			#cmd=$(echo 'sed -E "
			# /^(.{'$barcodelength'})(.{'${umilength}'})(.*)/ {
			# s/^(.{'$barcodelength'})(.{'${umilength}'})(.*)/\1\3/g
			# n
			# n
			# s/^(.{'$barcodelength'})(.{'${umilength}'})(.*)/\1\3/g
			# }" $convFile > ${crIN}/.temp\n
			# mv ${crIN}/.temp $convFile')
	if [[ $verbose ]]; then		if [[ $verbose ]]; then
	echo technology $technology		echo technology $technology
	echo barcode: $barcodelength		echo barcode: $barcodelength
	echo umi: $umilength		echo umi: $umilength
	echo $cmd		# echo $cmd
	fi		fi
	eval $cmd		#eval $cmd

	# add mock UMI (count reads instead of UMI) barcodelength=16, umi_default=10		# add mock UMI (count reads instead of UMI) barcodelength=16, umi_default=10
	perl sub/AddMockUMI.pl --fastq=${convR1} --out_dir $crIN --head_length=$barcodelength --umi_length=$umi_default		perl sub/AddMockUMI.pl --fastq=${convR1} --out_dir $crIN --head_length=$barcodelength --umi_length=$umi_default
	umilength=$umi_default		umilength=$umi_default
	umiadjust=0		umiadjust=0
	if [[ $chemistry == "SC3Pv3"]; then		if [[ $chemistry == "SC3Pv3" ]]; then
	chemistry="SC3Pv2"		chemistry="SC3Pv2"
	fi		fi

	#returns a combined R1 file with barcode and mock UMI		#returns a combined R1 file with barcode and mock UMI
	## barcode, 10 bp UMI, followed by TSO (if applicable)		## barcode, 10 bp UMI, followed by TSO (if applicable)
	mv $crIN/mock_UMI.fastq ${convR1}		mv $crIN/mock_UMI.fastq ${convR1}
			done
	fi		fi
	fi		fi

	#UMI
	echo " adjusting UMIs of R1 files"
	# check if original UMI is shorter than default
	if [[ 0 -gt $umiadjust ]]; then
	for convFile in "${convFiles[@]}"; do
	echo " handling $convFile ..."
	toS=`printf '%0.sA' $(seq 1 $(($umiadjust * -1)))`
	toQ=`printf '%0.sI' $(seq 1 $(($umiadjust * -1)))`
	#compute length of adjusted barcode + original UMI
	keeplength=`echo $((${barcode_default}+${umi_default}-($umiadjust * -1)))`
	#Add n characters to the end of the sequence
	sed -E "2~4s/(.{$keeplength})(.*)/\1$toS\2/" $convFile > ${crIN}/.temp
	mv ${crIN}/.temp $convFile
	#Add n characters to the end of the quality
	sed -E "4~4s/(.{$keeplength})(.*)/\1$toQ\2/" $convFile > ${crIN}/.temp
	mv ${crIN}/.temp $convFile
	echo " ${convFile} adjusted"
	done
	fi
	# check if original UMI is longer than default		# check if original UMI is longer than default
	if [[ 0 -lt $umiadjust ]]; then		if [[ 0 -lt $umiadjust ]]; then
	for convFile in "${convFiles[@]}"; do		for convFile in "${convFiles[@]}"; do

Admin message